The Impact Factors and Spatial Spillover of Industrial Green Development: Based on Cities in the Northwest Segment of the Silk Road Economic Belt

Abstract

With the increasing global emphasis on green development, industrial green development (IGD) has gradually gained attention as a crucial component of sustainable development. However, there is limited research evaluating and analyzing the IGD of cities with slow economic development and a high resource endowment. Utilizing empirical data gathered from cities in the northwest segment of the Silk Road Economic Belt spanning from 2009 to 2018, this research employs the entropy-weighted TOPSIS model to establish an indicator framework for evaluating IGD. Subsequently, this paper analyzes the impact of factors such as the economic foundation, innovation capacity and crucial guarantee on IGD, as well as the spatial spillover effects in cities of the northwest segment of the Silk Road Economic Belt using spatial panel data and the spatial Durbin model. The results indicate that the IGD levels of urban clusters exhibit the characteristics of plate-ladder-type difference, being “excellent in the southeast, good in the middle, and inferior in the northwest” displaying fluctuating upward trends and spatial clustering over time. Additionally, the degree of opening up also positively impacts IGD, while the level of economic development has a detrimental effect on IGD. Human capital and environmental regulations demonstrate notable spatial spillover effects on IGD. Our study enriches the evaluation system for IGD and provides recommendations for the industrial green transformation of cities along the Silk Road Economic Belt.

1. Introduction

In recent years, as China has emphasized high-quality economic development, an increasing number of cities have called for a greening of their economic growth. The Silk Road Economic Belt is a new economic development region formed by China in 2013, building upon the concept of the ancient Silk Road. Located inland, the cities along this belt possess abundant natural resources, contributing to rapid industrial development [1]. However, the rapid development of industry is inevitably accompanied by negative effects such as high energy consumption, pollution, inefficient production methods, ecological degradation and the depletion of natural resources. The concept of green development can be traced back to the 1960s when the American scholar, Boulding, published Stable Economics [2]. This concept has prompted a growing number of countries and institutions to focus on IGD. Germany proposed “GreenTech Made in Germany 2018: Umwelttechnik-Atlas für Deutschland” in 2018 [3] and the UK proposed the “Green Industrial Revolution Ten-Point Plan” in 2020 [4]. Industrial development, as a significant component of this growth, has become a crucial direction for promoting the transformation and upgrading of the industrial structure.

Consequently, the importance of industrial green development (IGD) for cities along the Silk Road Economic Belt has become increasingly apparent. Recognizing IGD as a key factor in achieving high-quality green development for cities along the Silk Road Economic Belt, the Ministry of Industry and Information Technology (MIIT) issued the “Industrial Green Development Plan (2016–2020)” in 2016 [5]. This plan emphasizes advancing supply-side structural reforms, promoting stable industrial growth and structural adjustment, advancing energy conservation and consumption reduction, achieving cost reduction and efficiency improvement, increasing the effective supply of green products and services, and addressing shortcomings in green development. Additionally, the Green Development Index System was also released in the same year [6]. However the resource-rich northwestern region of the Silk Road faces significant contradictions between economic development and industrial greenization due to its high dependence on industrial development [7]. This could lead to a lack of applicability of the nationwide indicator system in the northwest region. Therefore, without a suitable decision-making foundation, it is challenging to assess the IGD of cities.

Against this background, this research work selects industrial resource efficiency, industrial environmental governance, and industrial green growth to measure the level of IGD based on data from cities in the northwest segment of the Silk Road Economic Belt from 2009 to 2018. Then, variables from three dimensions, namely the economic foundation, innovation capability, and essential safeguards, are selected to investigate the influencing factors and spatial spillover effects in the process of IGD among cities.

The main conclusions are as follows. The IGD of cities in the northwest segment of the Silk Road Economic Belt show a pattern of “southeast better, middle moderate, northwest inferior,” which mirrors with the overall landscape of IGD in China. The degree of openness to the outside world has a positive and significant impact on IGD in the cities of the northwest segment, while the economic development level has a negative impact. Furthermore, the influences of human capital and environmental regulation exhibit significant spatial spillover effects.

2. Literature Review

IGD, as a vital component of green development, has consistently been a crucial topic of focus in the academic community. Many research works have evaluated IGD levels by considering input and output, expected output and adverse output [8,9,10,11]. This differs from previous industrial development assessments that focused solely on expected output, as IGD also considers the environmental impact of adverse outputs such as emissions. Su et al. evaluated IGD by assessing green production, green products and green industries [12]. They used an expert scoring method to assess the level of IGD in various provinces of China from 2005 to 2010. Chen et al. used the AHP method to evaluate IGD based on four aspects: the industrial green output index, the industrial green efficiency index, the industrial green innovation index and the industrial green policy [13]. The results indicated typical regional differences in China’s IGD, with a staggered distribution from the east to the west. The eastern region showed a higher level of IGD, and factors such as technological progress and innovation significantly influenced IGD. Wang et al. [14] assessed the green development level of the Yangtze River Economic Belt based on 11 indicators, including natural resource utilization, pollution emissions and green growth, using the green GDP accounting system, a multi-index measurement system, and a comprehensive indicator system. Wu and Huang utilized the entropy-weighted TOPSIS model to assess IGD in the Yangtze River Economic Belt [15]. Their assessment considered resource utilization efficiency, environmental governance intensity, innovation-driven capacity and growth quality level. While the mentioned studies focused on IGD in eastern Chinese cities, this study is set against the backdrop of cities in the northwest segment of the Silk Road Economic Belt. It employs the entropy-weighted TOPSIS model to evaluate IGD by assessing 15 indicators across three dimensions: industrial resource efficiency, industrial environmental pollution and industrial green growth.

On the basis of evaluating IGD, many researchers have begun to consider the influencing factors of IGD. According to the selected perspectives, the influencing factors can be broadly categorized into four main types. The first category is natural endowment factors, such as geographical location differences and the resource endowment structure [16,17]. Zhang and Song [18] observed a fluctuating trend of “worse-better-worse” between IGD and resource endowment. According to the resource curse theory, regions or countries with resource endowment advantages may experience suppressed economic development. This phenomenon has been confirmed to be significant in the context of IGD and transformation [19,20]. Additionally, the same influencing factors may have different effects on cities located in different spatial positions. Existing research has explored the influencing factors of IGD in regions such as the Yangtze River Delta, Beijing–Tianjin–Hebei and the Yellow River Basin [21,22,23], but there is still a lack of research on cities in the western regions of China.

The second category is economic development factors, such as economic development level, opening-up level and financial development level [24,25,26]. Chen et al., using the SDM, analyzed factors such as economic development and foreign direct investment in 30 provinces in China [27]. They found that the level of economic development has significant direct and spatial spillover effects on the IGD of these thirty provinces. Xu et al. conducted a quantitative analysis of the impact of finance on IGD using provincial-level data from China for the years 2006–2017 [28]. They found that the level of financial development has a positive impact on green development. However, this impact varies across regions, with a more significant influence in the eastern regions and a less pronounced effect as one moves westward.

The third category is innovation capacity, including innovative human capital and the level of technological innovation. Green innovation can have a positive impact on the green growth of the economy [29]. Li et al., utilizing panel data from 2005 to 2015 for 30 provinces in China, found an inverted U-shaped relationship between the regional IGD levels and the role of innovation in IGD [30]. They observed that innovation is most effective in low-carbon production, followed by resource reduction. However, Hu et al. conducted an examination of the innovation capabilities and IGD for 30 provinces in China from 2008 to 2017 [31]. Their results indicated that innovation capabilities exhibit a positive impact and vary across different regions. Fu et al., analyzing data from Chinese A-share-listed manufacturing companies from 2011 to 2017, also reached similar conclusions [32]. The perspectives of scholars on the impact of technological innovation on IGD appear to be inconsistent. Moreover, existing research has mainly analyzed samples at the provincial level, with a relatively coarse granularity.

The fourth category is government intervention, including environmental regulation and infrastructure construction [29,33,34,35]. Environmental regulation refers to the government’s formulation of relevant environmental policies for the control of environmental pollution. These policies can take various forms, including government regulatory measures, guiding social forces and harnessing market forces [36]. Existing studies indicate that the impact of environmental regulation on IGD is not always positive. In this context, regulatory policies with economic incentives are more conducive to IGD compared to administrative orders [37]. Additionally, policies that enhance public participation can improve the efficiency of IGD [38]. Du et al. analyzed the impact of environmental regulation based on panel data from 105 environmental monitoring cities in China [39]. The results suggest that when the level of economic development is low, environmental regulation inhibits the development of green technology innovation, thereby inhibiting industrial green development. However, when the level of economic development is higher, environmental regulation significantly promotes green technology innovation and industrial structural upgrading. Li et al. suggested that the government can enhance market concentration and establish entry barriers for high-pollution emission industries through environmental regulation [40]. This approach aims to improve the overall green productivity of various industries. Wang et al. conducted an analysis on panel data from industrial sectors in OECD countries, and their findings indicate that appropriately stringent environmental supervision systems are more conducive to green productivity, while excessively high or low levels are detrimental to IGD [41]. Wang et al., based on panel data from 268 prefecture-level cities in China from 2011 to 2017, conducted an analysis of the impact of transportation infrastructure on the efficiency of urban green development [42]. The results indicated that the increase in transportation infrastructure significantly enhanced the efficiency of urban green development, and this effect was more pronounced in larger cities. All three types of factors can have direct or indirect impacts on IGD. Therefore, in this study analyzing cities in the northwest segment of the Silk Road Economic Belt, we comprehensively consider these three categories of factors and select seven influencing factors for analysis.

3. Construction of Measurement Model and IGD Indicator System

Compared to the assessment of industrial development, the evaluation of IGD further delves into the green development aspects of cities. A quantitative analysis of the literature indicates that existing studies assess numerous indicators such as the industrial input–output, environmental pollution, green innovation and green efficiency. In order to objectively evaluate potential uncertainties, this paper adopts the entropy-weighted TOPSIS model to assess IGD in cities in the northwest segment of the Silk Road Economic Belt. The TOPSIS model, utilizing the positive and negative ideal solutions of comprehensive attribute problems, ranks the evaluation objects, while the entropy weight method is employed to calculate the weights. The TOPSIS model is adept at handling uncertainties and conducting systematic analyses. The model construction is informed by the research paradigm of Yang et al. [20], considering determinate and indeterminate factors as mutually influential and convertible under certain conditions, where the relevance expression can depict the relationship between the local and the overall context.

3.1. Evaluation Model Construction

Suppose there are $\mathrm{m}$ rating objects, and each evaluation object has$\mathrm{n}$ evaluation indicators. Set $X={\left(x_{ij}\right)}_{m\times n}$ as a matrix composed of indicators in each city and region, where $x_{ij}$ represents the $j\left(j=1,2,\cdots ,n\right)$ indicator of the $i\left(i=1,2,\cdots ,m\right)$ city. Note $x_j$ as the j column of matrix $X$. The steps of comprehensive evaluation with entropy-weighted TOPSIS are as follows:

(i)

Dimensionless data

In order to eliminate the influence provided by heteroscedasticity, dimensionless processing of data is required before data analysis. The specific operations are as follows:

The indicator for the bigger the better is as follows:

$$y_{ij=\left(x_{ij}-min{x}_j\right)/\left({maxx}_j-min{x}_j\right)}$$

(1)

The indicator for the smaller the better is as follows:

$$y_{ij}=\left({maxx}_j-x_{ij}\right)/\left({maxx}_j-{minx}_j\right)$$

(2)

(ii)

Entropy weight calculation

In practice, the magnitude of entropy is related to the variation degree of an indicator in the comprehensive evaluation system. The greater the of variation is, the smaller the entropy value will be; the more abundant the index information is, the greater its weight will be; and vice versa. Therefore, the weight of each index can be calculated according to the entropy reflected by the variation degree of each index value, namely the entropy weight. The specific process is as follows:

Entropy calculation:

$$E_j=\raisebox{1ex}{${\rm E}\left(y_i\right)$}\!\left/ \!\raisebox{-1ex}{$lnm$}\right.,(j=1,2,\cdots ,n)$$

(3)

Indicator differentiation calculation:

$$d_j=1-E_j,(j=1,2,\cdots ,n)$$

(4)

Solving for the entropy weight:

$$W_j=\frac{d_j}{\sum _{i=1}^n{d}_i},(j=1,2,\cdots ,n)$$

(5)

(iii)

Constructing the weighted normalization matrix

Based on the entropy weights obtained by solving in (2), the normalized data in Equation (1) are weighted, and the following matrix is constructed:

$$V={(v_{ij})}_{m\times n}=\left[\begin{array}{cccc}{W}_1{y}_{11}& W_2{y}_{12}& \cdots & W_n{y}_{1n}\\ W_1{y}_{21}& W_2{y}_{22}& \cdots & W_n{y}_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ W_1{y}_{m1}& W_2{y}_{m2}& \cdots & W_n{y}_{mn}\end{array}\right]$$

(6)

(1)

Determining the positive $V^{+}$ and negative $V^{-}$ ideal solutions of the evaluation object

The positive ideal solution is the hypothetical optimal solution, and each indicator achieves the best value among the alternatives; the negative ideal solution is the hypothetical worst solution, and each indicator in the negative ideal solution is the worst value among the alternatives. According to the weighted normalization matrix $V={\left(V_{ij}\right)}_{m\times n}$ constructed in (3), the positive ideal solution $V^{+}$ and the negative ideal solution $V^{-}$ are defined as follows:

$$\begin{array}{l}{V}^{+}={(v_j^{+})}_{j\in J}=\left\{(\max v_{ij}|j\in J)|i=1,2,\cdots ,m\right\}\\ V^{-}={(v_j^{-})}_{j\in J}=\left\{(\min v_{ij}|j\in J)|i=1,2,\cdots ,m\right\}\end{array}$$

(7)

where $J$ is the set of indicators after standard quantification.

(2)

Distance calculation

Based on the weighted normalization matrix $V={\left(V_{ij}\right)}_{m\times n}$ constructed in (3) and the positive ideal solution $V^{+}={\left({V_j}^{+}\right)}_{j\in J}$ and negative ideal solution $V^{-}={\left({V_j}^{-}\right)}_{j\in J}$ determined in step (4), the distance between the evaluation object and the positive ideal solution ${d_i}^{+}$ and the distance between the evaluation object and the negative ideal solution ${d_i}^{-}$ are defined as follows:

$$\left\{\begin{array}{c}{d}_i^{+}={\displaystyle \sum _{j=1}^n{(v_{ij}-v_j^{+})}^2}\\ d_i^{-}={\displaystyle \sum _{j=1}^n{(v_{ij}-v_j^{-})}^2}\end{array}\right.$$

(8)

where $i=1,2,\cdots ,\mathrm{m}$

(3)

Calculating the relative closeness of the evaluation object to the ideal solution, namely $C_i$

Based on the distance ${d_i}^{+}$ and ${d_i}^{-}$ between the evaluation object and the ideal solution given in (7), the relative closeness of the ideal solution of the evaluation object $C_i$ is calculated as follows:

$$C_i=\frac{{d_i}^{-}}{{d_i}^{+}+{d_i}^{-}},(i=1,2,\cdots ,m)$$

(9)

The magnitude of the relative closeness reflects the degree of merit of the evaluation object. The larger $C_i$ is, the closer the evaluation object is to the optimal level.

3.2. Indicator System Construction

Based on the theory of sustainable development of ecology, society and the economy, combined with the theory of industrial ecology and economics, industrial green development seeks to improve the efficiency of the use of industrial resources, reduce waste emissions in the production process and increase industrial green environmental governance and promote industrial green growth. Greadel [43] believes that improving the effective use of resources and reducing emissions of industrial waste in industrial production processes is an important part of industrial green development. At the same time, “Made in China 2025” [44] proposes vigorously developing green and environmentally friendly industries and promoting the transformation of industry to green industry. This paper follows scientificity, feasibility and data availability, and considering the characteristics of IGD in cities in the northwest segment of the Silk Road Economic Belt, we construct an evaluation indicator system for the level of IGD. This system encompasses three dimensions—industrial resource efficiency, industrial environmental governance and industrial green growth—and is further broken down into seven aspects, resulting in a total of 15 indicators. Meanwhile, we annotate each indicator based on the positive or negative impact of each indicator on IGD and their entropy weights, as shown in

Table 1. Indicator system of the IGD level.

Dimension Layer	Feature Layer	Metric Layer	Metric Units	Metric Properties	Weight
Industrial Resource Efficiency	Energy Efficiency	Energy consumption per unit of industrial value added	Tons of standard coal/CNY 10,000	−	1.61%
	Water Efficiency	Water consumption per unit of industrial added value	m2/CNY 10,000	−	1.57%
	Land Efficiency	Land area per unit of industrial value added	hm2/CNY 100 million	−	1.60%
Industrial Environmental Management	Pollution Emission Intensity	Industrial wastewater emissions per unit of industrial value added	10,000 tons/CNY 100 million	−	1.84%
		Industrial sulfur dioxide emissions per unit of industrial value added	Tons/CNY 100 million	−	1.60%
		Industrial smoke (dust) emissions per unit of industrial value added	Tons/CNY 100 million	−	1.88%
		Industrial solid waste emissions per unit of industrial value added	Tons/CNY 100 million	−	4.42%
	Pollution Control Intensity	Investment in industrial pollution control as a proportion of industrial value added	%	+	5.23%
		Centralized treatment rate of industrial wastewater	%	+	5.36%
Industrial green growth	Industrial Green Growth Efficiency	Growth rate of industrial value added	%	+	4.50%
		Industrial labor productivity	%	+	2.59%
		Revenue profit margin of industrial enterprises above scale	%	+	11.10%
	Industrial Green Growth Potential	Full-time equivalent of R&D personnel in industrial enterprises Share of employees	%	+	15.80%
		The proportion of R&D expenses of industrial enterprises in sales revenue	%	+	9.79%
		Number of patent applications filed by industrial enterprises	Number of items	+	31.11%

In order to ensure the consistency of the statistical caliber of the data, the scope of the data selected in this paper comprises the data of a total of 51 cities in the five provinces and regions of the northwest section of the Silk Road Economic Belt from 2009 to 2018, which were obtained from the statistical yearbooks of each city from 2010 to 2019, the China City Statistical Yearbook (2010–2019), the China Environmental Statistical Yearbook, the National Economic and Social Development Bulletin of each city (2010–2019), the China Industry Yearbook (2010–2019), the China Financial Statistics Yearbook (2010–2019), the China Industrial Enterprises Database, the China Business Database, the EPS Database, the Guotaian Database, CRDS and the Wind Database, etc.

.

4. Empirical Results and Analysis

4.1. Spatial–Temporal Evolution Analysis of the Level of Urban IGD in the Northwest Section of the Silk Road Economic Belt

Due to the significant impact of provincial-level administrative divisions on the development of cities in China, we initially conducted a statistical analysis of IGD in 51 selected cities based on provincial administrative regions. We measured the relevant data for cities along the Silk Road Economic Belt in the northwest region from 2009 to 2018 via the entropy weight method. The results depict the situation of IGD in the five northwestern provinces over different periods, as illustrated in Figure 1.

The results show that the growth rate of industrial greening in Gansu Province was the fastest compared with the other four provinces in Northwest China, with an overall growth of 19.329%. Meanwhile, the average level of IGD in Qinghai Province increased from 0.378 in 2009 to 0.382 in 2018, with an overall growth rate of 1.058%, which was the slowest among the five provinces. Ningxia Hui Autonomous Prefecture (hereinafter referred to as “Ningxia”) and Xinjiang Uygur Autonomous Region (hereinafter referred to as “Xinjiang”) had similar growth rates of IGD from 2009 to 2018, with Ningxia’s growth rate being 1.105% and Xinjiang’s growth rate being 2.718%. Additionally, 2016 marked a turning point in the level of IGD in the five northwestern provinces. Notably, this coincided with the period when the Ministry of Industry and Information Technology (MIIT) released the “Industrial Green Development Plan (2016–2020)”. This suggests a significant impact of these policies on IGD during this time.

To compare the temporal changes in IGD among different cities, we utilized the Natural Breaks method in ArcGIS 10.7 to map the spatiotemporal evolution of IGD levels in cities throughout the northwest segment of the Silk Road Economic Belt in 2009 and 2018. The results are illustrated in Figure 2.

In the northwest segment of the Silk Road Economic Belt, the IGD levels in the majority of cities exhibited a significant increase from 2009 to 2018. In 2009, the IGD levels for various cities ranged between 0.239 and 0.618, while in 2018, the range expanded to 0.245 and 1.063. Spatially, city clusters demonstrated a plate-ladder-type differentiation with characteristics of “southeast superiority, middle excellence, and northwest inferiority”. Within the southeastern city cluster, the cities in Shaanxi province, except for Hanzhong, Yulin and Shangluo, either remained unchanged or experienced a declining trend in IGD levels from 2009 to 2018. The other cities in the region showed significant improvement. In Ningxia, apart from Wuzhong and Shizuishan, which experienced a decline (21.856% and 9.792%, respectively), the cities witnessed an upward trend. Cities in the middle section of Gansu, including Jinchang, Qingyang, Dingxi and Longnan, experienced a downward trend. Qinghai province saw a significant decrease, with Yushu Tibetan Autonomous Prefecture, Golog Tibetan Autonomous Prefecture, Haibei Tibetan Autonomous Prefecture and Haidong City exhibiting a “N”-shaped decline from west to east. In Xinjiang, regions such as Hami, Tacheng, Altay, Bortala Mongol Autonomous Prefecture and Changji Hui Autonomous Prefecture all showed a declining trend. Additionally, compared to cities in other provinces in the northwest segment of the Silk Road Economic Belt, the IGD growth rate in Xinjiang’s urban cluster was relatively slow.

From an overall spatial perspective, there was a noticeable improvement in IGD in the surrounding cities with Xi’an as the center from 2009 to 2018, forming a positive spatial clustering. However, it is worth noting that in Xinjiang and Qinghai Province, due to undertaking industrial transfers and considering high returns from effectively allocating resources and reducing production costs as development goals, there was a spatial aggregation of decreasing IGD. This suggests the presence of spatial clustering characteristics in the IGD levels in the northwest segment of the Silk Road Economic Belt.

To further compare the disparity in IGD among different cities, this study employs the kernel density estimation method to illustrate the dynamic evolution trends of IGD in the urban clusters along the northwest segment of the Silk Road Economic Belt for the years 2009, 2013 and 2018, as shown in Figure 3.

The center of the IGD kernel density curve shows a left-skewed trend, indicating a stable and concentrated area of IGD levels among the 51 prefecture-level administrative cities. Compared to 2009, the “peak” in 2018 is relatively gentle, with a slight decrease in the peak value, suggesting a gradual narrowing of the IGD level’s disparity among cities. Simultaneously, the right tail of the curve displays a gradual rise, indicating an increasing number of cities with IGD levels above 0.6, signaling positive development momentum.

4.2. The Influencing Factors of Urban IGD in the Northwest Section of the Silk Road Economic Belt

4.2.1. Variable Selection

Based on the assessment of IGD levels, this paper considers IGD as the dependent variable and examines the influencing factors from three dimensions: the economic foundation, innovation capability and essential guarantees. Firstly, indicators commonly used to measure the economic foundation include the economic development level, the financial development level, and openness to the outside world [25,27,28]. The economic development level is a crucial indicator measuring the comprehensive strength of regional economic development, characterized by per capita GDP. The dynamic evolution characteristics of environmental externalities brought about by economic development may impact IGD. The financial development level is a significant manifestation of the quality improvement of regional economic development, represented by the Digital Financial Inclusion Index. The continuous development of digital finance propels positive economic development in cities, thereby potentially influencing IGD levels indirectly. Openness to the outside world is a key strategic means of jointly promoting green economic development with neighboring countries, represented by the ratio of the volume of imports and exports to GDP. The construction of the Silk Road Economic Belt is a crucial pathway to open up domestic and international dual circulation. The urban clusters in the northwest region, benefiting from their geographical advantages, can better utilize the level of openness to promote regional IGD.

Innovation capability, by accelerating the industrial green transformation to enhance IGD, typically considers technological innovation and human capital [29,31]. Technological innovation brings about new technologies and products, subsequently altering the environment and industrial production methods through technological advancements, thereby promoting the sustainable development of the green economy. In this paper, the level of technological innovation is represented by the ratio of the number of granted patents to GDP. The accumulation of human capital is an inherent driving force for technological innovation, and the continuous promotion of technological talents promotes the speed of technological innovation. In this study, the proportion of full-time-equivalent R&D personnel to the annual average employment rate is used to represent the level of human capital.

Important guarantees are provided by the government, typically through two aspects: environmental regulations and infrastructure construction [39,42]. Firstly, effective environmental regulations play a crucial role in addressing environmental pollution, urging industrial enterprises to reduce carbon emissions and efficiently utilize resources and energy. In this paper, the ratio of pollution control investment to GDP is used to represent environmental regulations. On the other hand, infrastructure construction can indirectly promote talent aggregation, technological assistance and economic cooperation, thereby influencing IGD. Therefore, we use the ratio of per capita road area to represent the level of infrastructure construction. Detailed indicators and codes are shown in

Table 2. Influencing factors of urban IGD.

Influencing Level	Variable Code	Variable Meaning	Indicator Calculation
Economic basis	LNGDP	Economic development level	GDP per capita
	LNFIN	Financial development level	Digital Financial Inclusion Index
	LNOPEN	Degree of opening-up	Import and export quota/GDP
Innovation capacity	LNRL	Human capital level	Full-time equivalent of R&D personnel/annual average number of employees
	LNRD	Scientific and technological innovation	Patent authorization amount/GDP
Crucial Guarantee	LNINF	Infrastructure construction level	Road area per capita
	LNER	Environmental regulation	Pollution control investment amount/GDP

.

4.2.2. Model Specification

According to Tobler’s “First Law of Geography”, there is a correlation between phenomena, but the correlation is stronger among entities that are closer in proximity. Therefore, we process the spatial panel data and construct a least squares panel regression model and a spatial panel model to assess the basic effects and spatial effects, respectively.

(1)

OLS Panel Regression Models:

$$\begin{gathered} LN{GIDL}_{it}={\beta }_0+{\beta }_{1}LNPGD{P}_{it}+{\beta }_{2}LN{ER}_{it}+{\beta }_{3}LN{INF}_{it}+{\beta }_{4}LN{OPEN}_{it}+{\beta }_5{LNRD}_{it}+{\beta }_6{LNRL}_{it}+ \\ {\beta }_7{LNFIN}_{it}+{\epsilon }_{it} \end{gathered}$$

(10)

(2)

Space Panel Model:

$$\begin{array}{cc}\hfill LN{GIDL}_{it}={\beta }_0& +{\rho WLN{GIDL}_{it}+\beta }_{1}LNPGD{P}_{it}+{\beta }_{2}LN{ER}_{it}+{\beta }_{3}LN{INF}_{it}+{\beta }_{4}LN{OPEN}_{it}+{\beta }_5{LNRD}_{it}\hfill \\ & +{\beta }_6{LNRL}_{it}+{\beta }_7{LNFIN}_{it}+{\theta }_{1}LNPGD{P}_{it}+{\theta }_{2}LN{ER}_{it}+{\theta }_{3}LN{INF}_{it}+{\theta }_{4}LN{OPEN}_{it}\hfill \\ & +{\theta }_5{LNRD}_{it}+{\theta }_6{LNRL}_{it}+{\theta }_7{LNFIN}_{it}+{\epsilon }_{it}\hfill \end{array}$$

(11)

① Spatial Autoregressive Model (SAR):

$$Y_t=\alpha l_n+\delta W{Y}_t+X_t\beta +{\epsilon }_i$$

(12)

② Spatial Error Model (SEM):

$$Y_t=\alpha l_n+X_t\beta +{\mu }_t$$

(13)

$${\mu }_t=\lambda W_{{\mu }_t}+{\epsilon }_t$$

(14)

③ Spatial Dubin model (SDM):

$$Y_t=\alpha l_n+\delta W{Y}_t+X_t\beta +W{X}_t\theta +{\epsilon }_t$$

(15)

The measurement of spatial distance between regions is the premise of spatial econometric analysis. The spatial data from n regions is denoted as ${{\left\{X_i\right\}}_{i=1}}^n$, where the subscript $\mathrm{i}$ represents the region $\mathrm{i}$. The distance between the region $i$ and the region $j$ is marked as $W_{ij}$, then the “spatial weight matrix” can be defined as follows:

$$W=\left(\begin{array}{ccc}{W}_{11}& \cdots & W_{1n}\\ ⋮& \ddots & ⋮\\ W_{n1}& \cdots & W_{nn}\end{array}\right)$$

(16)

where the main diagonal element is $W_{11}=\cdots =W_{nn}=0$ (the same region distance is 0). Obviously, the spatial weight matrix $\mathrm{W}$ is symmetric. The most commonly used distance function is “adjacent”. When $W_{ij}=1$, region $\mathrm{i}$ shares a boundary with region $\mathrm{j}$; when $W_{ij}=0$, region $\mathrm{i}$ and region $\mathrm{j}$ have no common boundary.

According to the basic form of the spatial panel model, if $\rho =0$ and $\delta =0$, only the spatial error interaction effect exists using the spatial error model (SEM). If $\lambda =0$ and $\delta =0$, only the spatial endogenous interaction effect exists, using the spatial autoregressive model (SAR), namely the spatial lag model (SLM). If $\lambda =0$, there is both a spatial error interaction effect and a spatial endogenous interaction effect. In this case, the spatial Durbin model (SDM) is used.

In this paper, the adjacency relation is defined based on the distance between regions, and the adjacency spatial weight matrix is used, which is represented by $W_1$ in this paper. The matrix formula is shown in Equation (17):

$$W_{ij}=\left\{\begin{array}{c}1,Spatialunitsiandjareadjacent\\ 0,Spatialunitsiandjarenotadjacent\end{array}\right.$$

(17)

In this paper, the adjacency matrix, geographic distance matrix and economic geographic weight matrix are constructed, and the spatial correlation is tested using Stata 17MP software. Moran’s global spatial autocorrelation coefficient is used to reflect the spatial dependence of the IGD level among cities in the northwest section of the Silk Road Economic Belt. The calculation formula is as follows:

$$\mathrm{M}\mathrm{o}\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{’}\mathrm{s}\mathrm{I}=\frac{n\sum _{i=1}^n\sum _{j=1}^n{W}_{ij}(y_i-\overline{y})(y_j-\overline{y})}{\sum _{i=1}^n\sum _{j=1}^n{W}_{ij}\sum _{i=1}^n(y_j-\overline{y})}$$

(18)

where $n$ is the total number of prefecture-level cities; $y_i$ and $y_j$ represent the IGD level of city $i$ and city $j$, respectively; $\overline{y}$ is the average level of IGD in each two cities; and $W_{ij}$ is the spatial adjacency weight matrix. The value of Moran’s I is between −1 and 1. If the corresponding p value is statistically significant, it shows that the level of IGD in the sample cities has obvious spatial autocorrelation.

The results are shown in

Table 3. Moran’s I of IGD level in northwest cities from 2009 to 2018.

Year	Adjacency Matrix		Geographical Distance Matrix		Economic Distance Matrix
	Moran’s I	p	Moran’s I	p	Moran’s I	p
2009	0.1833	0.0000	0.2498	0.0000	0.5016	0.0231
2010	0.1613	0.0000	0.2019	0.0000	0.4150	0.0590
2011	0.1523	0.0000	0.2175	0.0000	0.4001	0.0684
2012	0.1636	0.0000	0.1852	0.0000	0.1387	0.4792
2013	0.1206	0.0000	0.1750	0.0000	0.1651	0.4151
2014	0.1162	0.0000	0.1600	0.0000	0.0115	0.8897
2015	0.1384	0.0000	0.1766	0.0000	0.1311	0.5074
2016	0.1326	0.0000	0.1586	0.0000	0.0985	0.6019
2017	0.1240	0.0000	0.1729	0.0000	0.1083	0.5718
2018	0.1725	0.0000	0.1742	0.0000	0.0501	0.7576

and except for the economic distance matrix, the global Moran’s I of the urban IGD level in the northwest section of the Silk Road Economic Belt using the adjacency matrix and the geographical distance matrix is significantly positive at the level of 1%. This shows that the IGD level of each city has significant spatial dependence. Therefore, it is more accurate to use the spatial panel model to study the influencing factors of the urban IGD level in the northwest section of the Silk Road Economic Belt than other general panel models.

5. Analysis of Empirical Test Results

5.1. Empirical Test on Influencing Factors of IGD in the Silk Road Economic Belt

Firstly, the data were processed and examined. All variables passed unit root tests, cointegration tests and multicollinearity tests. The LM test results indicate a significant rejection of the null hypothesis at a 1% statistical level, suggesting the selection of the spatial Durbin model (SDM) for a more optimal fit in the analysis. To explore the spatial effects and influencing factors of IGD levels in the northwest segment of the Silk Road Economic Belt, this study compares ordinary least squares (OLS), spatial lag model (SLM), spatial error model (SEM) and SDM regressions as shown in

Table 4. Results of the spatial model of the level of IGD in the northwest section of the urban agglomeration based on the adjacency matrix.

Variables	Model
	OLS	SAR	SEM	SDM
LNGDP	0.036 **	−0.06	−0.061 *	−0.068 *
	(2.27)	(−1.59)	(−1.67)	(−1.86)
LNER	−0.007	−0.032 *	−0.038 **	−0.039 **
	(−0.15)	(−1.89)	(−2.23)	(−2.24)
LNOPEN	−0.029 ***	0.01	0.011 *	0.010 *
	(−5.81)	(1.64)	(1.94)	(1.73)
LNINF	−0.114	0.178	0.220 *	0.192
	(−0.34)	(1.36)	(1.7)	(1.44)
LNRD	0.108 ***	0.013	0.016 **	0.009
	(9.26)	(1.64)	(2.05)	(1.07)
LNRL	0.003	0.007	0.009 *	0.008
	(0.2)	(1.32)	(1.95)	(1.61)
LNFIN	−0.018	0.003	0.005	0.002
	(−1.30)	(0.25)	(0.49)	(0.19)
W·LNGDP				−0.196
				(−1.00)
W·LNER				−0.233
				(−1.63)
W·LNOPEN				0.081 *
				(1.78)
W·LNINF				1.086
				(1.08)
W·LNRD				0.031
				(0.66)
W·LNRL				0.075 ***
				(2.65)
W·LNFIN				0.006
				(0.07)
λ			−0.940 ***
			(−3.85)
ρ		−0.689 ***		−1.195 ***
		(−3.35)		(−4.83)
sigma2		0.008 ***	0.008 ***	0.008 ***
		(15.78)	(15.55)	(15.38)
R2	0.179	0.006	0.003	0.033
Log−L		489.8233	492.1815	498.1248
N	510	510	510	510

Note: ***, **, and * indicate that the estimated coefficients are significant at the 1%, 5% and 10% levels, respectively, with the Z statistic in brackets.

.

The spatial lag coefficient (ρ) of the SAR model is −0.689, the spatial error coefficient (λ) of the SEM model is −0.940, and the lag coefficient of the SDM model is −1.195. All these coefficients are significant at a 1% level. This indicates a significant spatial clustering effect in the IGD levels among cities in the northwest segment of the Silk Road Economic Belt. To avoid the degeneration of the SDM into a spatial lag model and a spatial error model, LR tests are conducted to validate the assumptions of the two degenerate models. The LR test statistic, which is the difference in the log-likelihood values of the two models, follows a chi-square distribution with degrees of freedom K. The values for LR_spatial_lag and LR_spatial_error are 16.60 and 11.89, respectively. In the Wald test, the values for Wald_spatial_lag and Wald_spatial_error are 41.45 and 42.89, respectively. In both tests, the null hypothesis of SDM model degeneration is rejected at a 10% significance level, indicating that the spatial Durbin model is the optimal choice.

Secondly, using the Hausman test to assess the fit of fixed effects and random effects models to the data, the p-value is 0, and the chi-square value is 25.56. This implies a rejection of the null hypothesis at a 1% significance level, favoring the adoption of a fixed effects model. Therefore, this study chooses the spatial Durbin fixed effects model to analyze the influencing factors and spatial effects of IGD levels in cities in the northwest segment of the Silk Road Economic Belt. The direct, indirect and total effects of the spatial Durbin fixed effects model in the cities in the northwest segment of the Silk Road Economic Belt are presented in

Table 5. Spatial effects’ decomposition of the spatial Durbin model with dual fixed effects.

	Direct Effect	Indirect Effect	Total Effect
LNGDP	−0.062	−0.061	−0.123
	(−1.57)	(−0.66)	(−1.44)
LNER	−0.033 **	−0.087	−0.120 *
	(−1.99)	(−1.32)	(−1.82)
LNOPEN	0.008	0.035	0.043 **
	(1.36)	(1.59)	(2.06)
LNINF	0.168	0.38	0.548
	(1.28)	(0.81)	(1.17)
LNRD	0.008	0.01	0.018
	(0.92)	(0.43)	(0.85)
LNRL	0.006	0.031 **	0.037 ***
	(1.21)	(2.37)	(2.85)
LNFIN	0.002	0.004	0.006
	(0.15)	(0.11)	(0.15)

Note: ***, **, and * indicate that the estimated coefficients are significant at the 1%, 5%, and 10% levels, respectively.

.

We found that, for each unit increase in the economic development level, there is a significant negative impact on IGD, resulting in a decrease of 0.068 units at a 10% significance level. According to the principles of the environmental Kuznets curve, when a country or region has a lower economic development level, environmental pollution is relatively light. As per capita income increases, environmental pollution gradually rises. The northwest segment of the Silk Road Economic Belt, being in the western part of China, experiences relatively slow economic development, with a predominant economic structure based on resource-intensive industries and traditional agriculture. This sluggish economic development places the cities in the northwest segment at the left end of the curve’s “turning point,” indicating that economic development inhibits IGD. The level of financial development has a nonsignificant positive impact on IGD, and there is no spatial effect. This suggests that the current level of financial development in the northwest region lags behind the overall level in the eastern coastal regions, and thus it does not play a positive role in promoting IGD. For each unit increase in the level of openness, IGD increases by 0.010 units, exhibiting a significant positive impact at a 10% significance level. Additionally, the degree of openness has a significant spatial effect. This indicates that cities in the northwest segment of the Silk Road Economic Belt, with a significant geographical advantage in terms of openness, experience mutual influence, promoting industrial transfer and agglomeration, optimizing the regional industrial green layout and thereby driving the IGD of city clusters.

In terms of innovation capability, the level of human capital exhibits a significant spatial spillover effect on IGD. For each unit increase in human capital under spatial weighting, IGD increases by 0.075 units, showing a significant promoting effect on IGD at a 1% significance level. Moreover, in SDM, human capital has a spatial spillover effect at a 5% significance level, indicating that an increase of 1 unit in human capital in neighboring cities leads to an increase of 0.031 units in IGD in the city. The new theory of economic growth suggests that enhancing human capital helps to improve labor productivity and the efficient use of resources. The accumulation of human capital is a core element for industrial enterprises to enhance their market competitiveness. The introduction of high-quality talents helps enterprises to improve their technological innovation levels, achieve maximum benefits, promote IGD, coordinate the relationship between industrial economic scale and ecological protection, and ultimately to achieve the goal of high-quality development of the industrial economy. On the other hand, the impact of technological innovation on IGD is not significant. One possible reason for this is that in the process of industrial transformation in the cities of the northwest segment of the Silk Road Economic Belt, traditional high-energy and high-pollution industries occupy a large amount of the technological innovation resources, which, to some extent, fails to demonstrate industrial transformation’s proactive role in promoting the transformation towards industrial greenization. Additionally, due to the insufficient original accumulation of technological innovation capital in cities in the northwest segment of the Silk Road Economic Belt, the level of technological innovation is relatively low, and the achievements of industrial innovation have not been widely distributed, which does not actively promote the upgrading of urban industrial structure and industrial transformation.

In terms of crucial safeguards, for each unit increase in environmental regulation, there is a significant negative impact on IGD, leading to a decrease of 0.039 units at a 5% significance level, with a notable feedback effect. This indicates that stringent environmental policies in the northwestern cities with relatively lower levels of economic development may paradoxically result in industrial pollution exceeding the standards, thereby causing a lower level of IGD. Meanwhile, the spatial direct effect of environmental regulation on urban IGD levels shows a significant negative impact due to the mutual influence among neighboring cities in adopting environmental regulations. The level of infrastructure construction does not exhibit a significant impact on the industrial green development level of cities.

5.2. Robustness Test

To test the robustness of the above conclusions, we conduct tests by changing variables and transforming the spatial matrix. The total GDP was used as a replacement for per capita GDP to represent the level of economic development. Additionally, the geographic distance matrix was used as a replacement for the adjacency matrix, as shown in

Table 6. Robustness test results.

Variables	SDM
LNGDP	−0.001	W·LNGDP	0.008
	(−0.25)		−0.81
LNER	−0.026	W·LNER	−0.034
	(−1.50)		(−0.91)
LNOPEN	0.008	W·LNOPEN	−0.011
	(1.26)		(−0.78)
LNINF	0.132	W·LNINF	0.167
	(0.97)		(0.58)
LNRD	0.008	W·LNRD	0.038 **
	(0.98)		(2.15)
LNRL	0.006	W·LNRL	0.004
	(1.08)		(0.38)
LNFIN	0.004	W·LNFIN	0.016
	(0.42)		(0.76)
rho	−0.123 *	N	510
	(−1.75)	R-squared	0.077
sigma2_e	0.009 ***
	(15.93)

Note: ***, **, and * indicate that the estimated coefficients are significant at the 1%, 5% and 10% levels, respectively.

. The coefficients of the variables remained consistent with the previous results, demonstrating the robustness of the research findings.

6. Conclusions and Policy Recommendations

6.1. Conclusions

This study focuses on the 51 prefecture-level administrative cities in the northwest segment of the Silk Road Economic Belt from 2009 to 2018. It evaluates the cities’ IGD using the entropy-weighted TOPSIS model from three dimensions: industrial resource efficiency, industrial environmental governance. Subsequently, this research employs the Spatial Durbin Fixed Effects Model to analyze the influencing factors and spatial effects of IGD in cities in the northwest segment of the Silk Road Economic Belt, considering 15 indicators across three dimensions: the economic foundation, innovation capability, and critical support. The conclusions drawn are as follows:

The overall IGD level of cities in the northwest section of the Silk Road Economic Belt has significantly improved, exhibiting a plate-ladder-type differentiation feature of “southeast better, middle good, northwest worse”. Overall, there has been a shift from local to overall concentration of IGD, with most cities showing spatial concentration in 2018, indicating the distinct characteristics of industrial agglomeration. The level of economic development has a significant inhibitory effect without spatial effects, and the degree of openness has a significant promoting effect with regional spatial effects. In the innovation capability dimension, human capital has a significant positive spatial spillover effect. In the critical support dimension, environmental regulation significantly inhibits industrial green development and has a significant positive spatial feedback effect.

6.2. Policy Recommendations

Based on the conclusions of this study, we propose the following three policy recommendations. The first is to enhance the level of open platforms, promoting shared development in the industry. This includes making full use of high-quality open platforms, enhancing the specialization, internationalization and market-oriented characteristics of economic and trade activities, and promoting the opening of the ecological industry chain. Additionally, there is a focus on enhancing the external opening of border cities. Taking examples such as the Khorgos Port and Alataw Pass into account, new port-type cities should be developed to enhance the capacity for opening up in a more extensive manner.

Secondly, the government should stimulate the vitality of scientific research and innovation and improve the rate of industrial green transformation. This includes fully leveraging the driving role of provincial capital cities to propel the collaborative development of surrounding cities, actively nurturing technology-based enterprises and comprehensively implementing a gradient cultivation project for innovative enterprises; encouraging the establishment of technology-based small and medium-sized enterprises in northwest cities, and supporting their integration into the national industrial innovation network; and improving the supply quality of innovation entities, including universities and research institutions in the northwest region, optimizing the urban higher education and research system, and consequently attracting more innovative talents.

Finally, we recommend optimizing environmental regulation methods and strengthening industrial green institutional constraints. Environmental regulation should be differentiated according to the degree of difference in the level of regional IGD. For areas with a low level of IGD, such as Qinghai, Ningxia and Xinjiang, a reasonable environmental regulation mechanism should be formulated from the source. For regions with moderate IGD levels such as Gansu, it is advisable to formulate environmental regulations that align with local requirements. This involves guiding capital towards environmentally friendly industrial enterprises. For regions with higher levels of IGD, such as Shaanxi Province, it is recommended to formulate environmental regulations that encourage the research on and development of green technologies.

Then, the intensity of environmental regulation is reasonably set. In pollution-intensive industrial areas, complementary measures such as green technology innovation and industrial structure adjustment should be implemented. This involves increasing research and development efforts, integrating resource elements and initiating environmental regulation construction from the source. In areas with moderate- to light-polluting industries, it is recommended to increase environmental supervision efforts, promote energy conservation and emission reduction in industrial enterprises and strictly control pollutant emissions, making environmental supervision a core aspect.