The impact of market integration on industrial green transformation: A case study of the yangtze river delta in China

Abstract

Regional integration plays an important role in dicarbon reduction and sustainability development. Based on dynamic panel model and spatial econometric model, this study analyzes the impact of market integration (MI) on industrial green transformation (IGT). The study finds that: (1) MI has a nonlinear relationship with the IGT. With the increasing of MI degree, the IGT shows the characteristic of first rising and then declining. (2) MI plays a role in the IGT mainly through industrial agglomeration, resource mismatch and green technological progress, respectively. (3) The impact of MI on IGT exhibits spatial heterogeneity. Although MI significantly promotes IGT in core cities, its impact on the IGT in peripheral cities is not significant. (4) The IGT has spatial spillover effect, and the improvement of MI degree in adjacent regions can also promote the IGT in the cities.

1. Introduction

In the background of decentralization, local governments have achieved rapid economic growth in their jurisdictions by vigorously developing the manufacturing and real estate industries. However, problems such as serious market segmentation (MS) and vicious competition have also emerged, limiting the integrated development of city-region [1]. In 2022, the Chinese government explicitly proposed to establish a national unified big market. This means increasing the strength of the fight against MS and local protection, breaking down the economy of administrative districts, and achieving the free flow of commodities and factors across regions. As the foundation of the national unified market, MI has an important impact on promoting the flow of resources and realizing the optimization industrial structure [2,3]. However, scholars have preliminarily analyzed the relationship between MI or MS on regional economic growth, economic disparity [4], and pollution emission [5], respectively. They pay less attention to the role of MI on IGT, making it difficult to comprehensively reveal the resource-environmental effects of regional integration [6].

With the deepening of economic globalization and regional integration, some scholars have explored the effects of regional integration on industrial transformation and upgrading from the perspectives of trade liberalization [7]. Baldwin and Okubo (2006) finds that regional economic integration reduces trade costs and promotes cross-regional relocation of enterprises and industrial transformation and upgrading [8]. In recent years, regional integration has gradually become an important national development strategy, and its socioeconomic impact within countries has become a focus [9,10]. Chen (2016) finds that regional integration promotes inter-regional technology sharing and green innovation, improves energy utilization efficiency and promotes IGT [11]. In addition, Chen et al. (2021) analyzed the impact of Yangtze River Delta (YRD) integration on efficiency of industrial green development, and found that regional integration improved the efficiency of IGT [12].

In general, scholars mainly analyze the role of regional integration on the transfer of polluting industries, but lack of in-depth exploration on the effect of regional integration on the IGT. These studies provide some support for exploring the low-carbon, green and sustainable development of the industry. However, the IGT is a dynamic and long-term process, and the impact of regional integration is affected by a combination of factors. Therefore, causal analysis and mechanism explanation based on a particular perspective cannot be convincing. This paper tries to make up for the deficiencies in the existing research. The innovations are in the following aspects: (1) in the theoretical aspect, this paper illustrates the micro-action mechanism of MI on IGT from three dimensions, including scale, structure, and technology. This enriches the international trade theory and the industrial transfer theory. (2) In the index measurement, this study adopts the SBM (slacks-based measure) model to measure green total factor productivity (GTFP). Compared with the previous single-dimension indicators focusing on the environment or industrial structure, this measure can more accurately assess the input-output efficiency of the industry, and provide novel ideas for understanding the IGT; (3) In the empirical methodology, this paper takes into account the possible existence of industrial or regional heterogeneity, as well as spatial spillover effects, and adopts appropriate econometric methods to test them. The empirical results can deepen the understanding of the strategies and performance of IGT effect of MI.

2. Literature review

2.1. Meaning and measurement of IGT

At present, researchers mainly define the IGT from the structural and performance perspectives. From the structural perspective, Li et al. (2011) interpret the IGT as “in order to realize the greening of the industrial structure, a new industrialization guided by the orientation of intensive resource utilization and environmental friendliness is implemented” [13]. From the perspective of transformation performance, Peng (2021) explain the IGT as “the process of intensive utilization of resources, reduction of total pollution emission and pollution emission intensity, reduction of production damage and improvement of production efficiency” [14]. Although scholars have defined the concept and connotation of IGT, they have not yet formed a unified standard for its measurement, and have formed the following three approaches. Based on the connotation of IGT, Liu et al. (2021) measure IGT by the ratio of the output value of polluting industries in the industrial sector. As the connotation of IGT is more complex, scholars have tried to construct a comprehensive indicator system to measure IGT [15]. For example, Chen et al. (2016) constructed a comprehensive indicator system from green output, efficiency, innovation to measure IGT. In recent years, scholars have used input-output analysis to quantitatively evaluate the IGT by calculating the GTFP [16]. Gong et al. (2020) used the super-efficient SBM model to measure the GTFP of 27 subsectors in China's industry [17].

2.2. Influence mechanism of MI on IGT

MI is an important dimension of regional integration. Through industrial agglomeration, factor flow and knowledge spillover, respectively, MI affects the spatial distribution of industries, resource allocation efficiency and green technology progress, thereby promoting the IGT (Fig. 1). Specifically, the influence mechanism is mainly manifested in the following aspects.

Fig. 1.

Mechanisms of MI influencing IGT.

2.2.1. MI, industrial agglomeration and IGT

The MI reduces the cost of enterprise migration across regions or industries, promotes inter-regional or inter-industry industrial transfer or agglomeration, and thus promotes the IGT. Firstly, with the improvement of MI, isolated markets between regions form a unified regional big market and promote industrial agglomeration through local market effect [18]. In order to capture markets and reduce transportation costs, firms tend to cluster in regions with larger market sizes to reap the benefits of agglomeration economies [19]. For example, it reduces the shared costs of polluting facilities and their treatment costs [20]. Secondly, with trade facilitation and factor flow liberalization, the cost of enterprise relocation across regions or industries is reduced, leading to the transfer of pollution-intensive industries across regions, and reducing the proportion of regional pollution-intensive industries [21]. Finally, although MI provides the benefits of agglomeration economy for industrial transfer or agglomeration, excessive industrial agglomeration can also lead to negative externalities [22]. In particular, excessive industrial agglomeration raises the price of factors of production, resulting crowding-out effect. This reduces the efficiency of resource and energy utilization, thus hindering IGT [23].

2.2.2. MI, resource allocation efficiency and IGT

The MI facilitates the free flow of factors, improves the efficiency of resource allocation. This helps to give full play to regional comparative advantages and reduce product costs, improves manufacturing efficiency, and ultimately promotes IGT. Firstly, with the improvement of MI level, this means that under the dual effects of free flow of production factors and competition mechanisms, the allocation of resources and energy is directed towards regions or industries with high marginal returns. This not only improves the efficiency of resource allocation, but also promotes IGT [24]. Secondly, the free flow of factors can help alleviate the distortion of factor market prices, and improve the efficiency of the use of factors, thus improving the productivity of enterprises and reducing production costs [25]. However, Que et al. (2018) found that distorted markets and their resulting MS produce more polluting emissions, reduce environmental efficiency, and thus hinder the IGT [26]. Finally, with the free flow of factors and the improvement of their allocation efficiency, some backward pollution-intensive enterprises are crowded out, reducing pollution emissions and promoting the IGT [27].

2.2.3. MI, green technological progress and IGT

The MI improves the level of urban green technology by enhanced knowledge spillover in order to promote the IGT. Firstly, with the development of MI, by the mechanisms of sharing, matching and learning, the industrial transfer and agglomeration of inter-cities accelerate the spatial reconfiguration of the industry. This promotes mutual learning and knowledge spillover, and generates technological innovation effects, leading to the development of IGT [28]. Secondly, MI increases inter-regional socio-economic linkages and generates spatial spillover effect, which also leads to the formation of regional innovation network. This is conducive to knowledge spillover and green technology level improvement, and provides technical support for IGT [29]. Finally, MI provides transaction environment for the cultivation of cross-regional technology market, reduces the transaction cost of green technology. This is conducive to the application of emission reduction technologies, and promotes inter-regional production technology spillover, thus providing technical support for IGT [30].

3. Indicator calculation and model construction

3.1. MI measurement

Drawing on the relative price method used in existing research [31], the level of regional integration is estimated by the relative difference in commodity prices between cities. Based on the information of cities, time, and commodity prices, the relative price method measures the MI level by calculating the relative price difference of commodities between different cities. In this study, the sample includes 41 cities in the YRD,1 and the time period is mainly taken from 2000 to 2020, and eight important commodities such as food, beverages, cigarettes, alcohol, and clothing, shoes, and hats are selected. This selection is on basis of the statistical description of the National Bureau of Statistics (NBS). The NBS statistical survey covers the prices of goods and services in 8 major categories and 268 basic classifications. The price indices of 8 major categories are published in the National Economic Bulletin every year.

This paper adopts this caliber of data to calculate the MI index. Firstly, pair the 41 cities and the price indices of eight commodities in these cities are first-order differenced to obtain ${\Delta Q}_{ijt}^k\text{,}$ and then its absolute value $\mid {\Delta Q}_{ijt}^k\mid$ is taken.

$${\Delta Q}_{ijt}^k=\mathit{ln}\left(\frac{P_{it}^k}{P_{it‐1}^k}\right)‐\mathit{ln}\left(\frac{P_{it}^k}{P_{jt‐1}^k}\right)=\mathit{ln}\left(\frac{P_{it}^k}{P_{it}^k}\right)‐\mathit{ln}\left(\frac{P_{it‐1}^k}{P_{jt‐1}^k}\right)$$ (1)

In equation (1), P denotes the retail price index of commodities, and the absolute value is taken mainly to alleviate the problem of the order of the city samples, and then the logarithmic number theory is taken on it to avoid the problem of heteroskedasticity and bias type of the data.

Secondly, according to the 20-year retail price index of commodities in 8 categories of commodities in 41 cities, the relative price in the form of difference between cities is obtained. Finally, Equation (2) is utilized to eliminate the non-additivity problem due to heterogeneity associated with specific commodities, and the de-mean method is utilized to eliminate the price volatility effect associated with the characteristics of the commodities themselves.

$$\left|{\Delta Q}_{ijt}^k\right|-\left|\overline{{\Delta Q}_t^k}\right|=\left|a^k-a^{-k}\right|+\left|{\epsilon }_{ijt}^k-{\epsilon }_{ijt}^{-k}\right|$$ (2)

In equation (2), $a^k$ is mainly determined by the type of commodity and represents the bias due to individual fixed effects. ${\epsilon }_{ijt}^k$ denotes the price difference caused by different external conditions in each city. Given year t, the price index k of commodity $\mid {\Delta Q}_{ijt}^k\mid$, find the mean value of inter-city pairing $\mid Q_t^k\mid$, then subtract the mean value $\mid Q_t^k\mid$ from the difference price $\mid {\Delta Q}_{ijt}^k\mid$ between the cities of the cohort, and judge the fluctuation of prices in different cities by the variance of $q_{ijt}^k$, and the larger the value of Var($q_{ijt}^k$), it indicates that the larger the value of Var ($q_{ijt}^k$).

3.2. Measurement of green transformation of manufacturing industry

As the connotation of IGT is relatively rich, scholars have not yet formed a consensus on its concept, resulting in the lack of a unified standard for its measurement. In view of the subjectivity of indicator selection and weight assignment in the single indicator and comprehensive indicator system method, this study mainly utilizes the super-efficient SBM (slacks-based measure) model to measure the GTFP of manufacturing industry. (Table 1). The SBM model (slacks-based measure) can be used to deal with efficiency measurement problems with multiple input and output variables. The efficiency of green development of manufacturing industry is the result of a three-way trade-off between inputs (production factors), desired outputs (economic outputs) and non-desired outputs (environmental pollution). More directly, higher GTFP implies lower factor inputs and higher economic returns on the one hand, and lower pollution emissions on the other. This is highly consistent with the core of the IGT.

Table 1.

Measurement of green transformation in industry.

Type	Indicator	Definition
Input	Labor	Number of employees in manufacturing industry
	Capital	Investment in fixed assets of manufacturing industry
	Energy	Social electricity consumption in manufacturing industry
Desired output	Economic output	Manufacturing value added
Non-desired output	Environmental pollution	Manufacturing industry sulfur dioxide emissions
		Manufacturing wastewater emissions
		Manufacturing smoke emissions

3.3. Model construction

As a complex and dynamic process, the IGT is not only affected by MI, but also constrained by other factors [32]. At the same time, due to the phenomena of “inertia” and “path dependence” of economic activities, the current development state of individuals is often closely related to the past state. Therefore, when analyzing the inertia and path dependence of IGT development, it is necessary to consider overcoming deviations caused by historical factors. Compared with static panel models, dynamic panel models incorporate and control the lagged terms of the explanatory variables into the model, which more comprehensively examines the dynamic development process of economic activities and their path-dependence characteristics [33]. However, the accuracy of the estimation results is affected by a certain endogeneity between the lagged terms of the explanatory variables and the error terms. The dynamic panel GMM model takes the lagged variables as instrumental variables for estimation, which better overcomes the endogeneity problem between the lagged and error terms of the explanatory variables, and improves the accuracy of the estimation results [34].

$${\text{lnGTFP}}_{\text{it}}={\mathrm{a}}_0+{\mathrm{a}}_1{\text{MS}}_{\text{it}}+{\mathrm{a}}_2{\text{ER}}_{\text{it}}+{{\mathrm{a}}_3\mathrm{X}}_{\text{it}}+{\mathrm{u}}_{\text{it}}+{\mathrm{\epsilon }}_{\text{it}}$$ (3)

$${\text{lnGTFP}}_{\text{it}}={\mathrm{a}}_0+{\mathrm{a}}_1{\text{GTFP}}_{\text{it}‐1}+{\mathrm{a}}_2{\text{MS}}_{\text{it}}+{\mathrm{a}}_3{\text{ER}}_{\text{it}}+{{\mathrm{a}}_4\mathrm{X}}_{\text{it}}+{\mathrm{u}}_{\text{it}}+{\mathrm{\epsilon }}_{\text{it}}$$ (4)

Equation (3) represents the static panel model, and equation (4) represents the dynamic panel model, ${\text{lnGTFP}}_{\text{it}}$ denotes the green total factor of manufacturing industry; ${\text{MS}}_{\text{it}}$ denotes the degree of MS; ${\text{ER}}_{\text{it}}$ denotes the environmental regulation (ER); ${\mathrm{X}}_{\text{it}}$ denotes a series of control variables, such as the endowment of resources, agglomeration economy, and the level of openness to the outside world; ${\mathrm{u}}_{\text{it}}$ denotes the individual effect and is used to control the city-specific effect; ${\mathrm{\epsilon }}_{\text{it}}$ denotes the model error term; ${\mathrm{a}}_0$ denotes the intercept term (Table 2).

Table 2.

Main variable and their measurement methods.

Variable Type	Variable Name	Measurement Method
Dependent variable	GTFP	Manufacturing GTFP
Independent variables	Market segmentation(MS)	Measurement by the relative price index method
Control variables	Command environmental regulation	Industrial “three waste” pollutant removal rate
	Market environmental regulation	The proportion of investment in environmental protection to the total industrial output
	Voluntary environmental regulation	Number of environmental protection-related words in government work reports as a percentage of total words
	Resource endowment	Proportion of local employees in the extractive industry
	Agglomeration economy	Calculated based on formula (5)
	Industry structure	Industrial value added/GDP
	Energy consumption intensity	Electricity consumption in manufacturing industry/total manufacturing output
	GDP	Logarithm of GDP per capita
	FDI	Actual utilization of foreign capital/regional GDP
	Urban innovation	Logarithm of the number of urban patents per 10,000 people
	Urbanization rate	Non-farm population at the end of the year/total population
	City size	Logarithm of the number of urban population
	Transportation accessibility	Logarithm of road network density per capita

As there are more measurement indexes and calculation methods for agglomeration economy, this paper mainly adopts location entropy index to measure the level of industrial agglomeration to characterize the agglomeration economy by combining the data accessibility and research purpose.

$${LQ}_{ij}=\frac{q_{ij}}{\sum _{i=1}^n{q}_{ij}}/\frac{\sum _{j=1}^m{q}_{ij}}{\sum _{i=1}^n\sum _{j=1}^m{q}_{ij}}$$ (5)

In equation (5), ${LQ}_{ij}$ denotes the entropy of location; j represents industry; i represents city; $q_{ij}$ is the number of employment; $q_i$ denotes the number of employment in all industries; $q_j$ denotes the number of employment in the whole country, and q denotes the number of employment in all industries in the whole country.

4. Empirical result

4.1. Dynamic panel regression

The results show that the estimated coefficient of MS is negative, and the estimated coefficient of SMS is positive, indicating that there is a nonlinear relationship between MS and IGT (Table 3). Since the opposite of MS is MI, it indicates that as MI increases, the IGT shows a tendency to increase first and then decrease, the MI and IGT have an inverted U-shaped relationship. When the level of MI is low, the cost of inter-regional trade is higher. At this time, it is difficult for production factors to flow freely. This not only reduces the efficiency of resource allocation, but also hinders the diffusion of knowledge spillovers. Furthermore, it leads to convergence of industrial structure and low efficiency of enterprise resource utilization, thereby hindering IGT. As the degree of MI increases, the cost of commodity trade and the cost of factor flow are decreasing. This not only improves the efficiency of resource allocation and utilization, but also promotes industrial transfer and agglomeration of inter-regional or inter-industry. Furthermore, through the effect of resource allocation and agglomeration economy, IGT is promoted. The estimated coefficient of the quadratic term of MS is positive. The increase in MI indicates that it reduces the cost of enterprise relocation, and promotes industrial transfer and agglomeration of inter-regional. However, excessive industrial agglomeration leads to an increase in pollution emissions, exacerbating the consumption of resources and energy, resulting in negative effects and hindering the IGT. In addition, the estimated coefficient of the lagged term of GFTP is positive, this means that the IGT has obvious path-dependent characteristics and is affected by the industrial structure in the past.

Table 3.

Baseline regressions for MI affecting GTFP.

Variable	OLS	PD	DPD	GMM
MS	−0.052c (-1.10)	−0.03c (-1.24)	−0.012c (-2.39)	−0.015b (-2.53)
SMS	0.025c (1.03)	0.013b(1.00)	0.004b (0.36)	0.007b (0.50)
GTFPt-1			0.519a (8.42)	0.783a (15.29)
Command environmental regulation	0.001 (0.57)	0.004b (1.15)	0.003b (0.91)	0.004b (1.18)
Market environmental regulation	0.02b (8.45)	0.029b (1.82)	0.029b (3.09)	0.023a (2.55)
Voluntary environmental regulation	0.032 (0.39)	−0.091 (-1.58)	−0.087 (-1.52)	−0.081 (-1.41)
Agglomeration economy	0.016a (4.09)	0.022b (1.35)	0.009b (0.85)	0.013a (3.20)
Resource endowment	−0.002b (-2.33)	−0.003 (-0.47)	0.01c (0.04)	−0.02b (-0.54)
Industry structure	0.035 (1.54)	−0.184b (-2.63)	−0.095b (-2.47)	0.012b (0.73)
Energy consumption structure	0.025c (0.45)	−0.39a (-3.27)	−0.239b (-2.53)	−0.015b (-0.53)
GDP	0.015b (2.52)	0.027c (1.13)	0.009c (0.58)	0.017b (2.49)
FDI	0.005b (2.11)	0.003 (0.48)	0.12c (0.05)	0.013b (1.74)
Urban innovation	0.015a (4.50)	0.006b (0.83)	0.002a (0.37)	0.024a (1.31)
Urbanization rate	0.011 (0.56)	0.076 (2.41)	0.052 (2.37)	0.027 (2.10)
City size	0.032 (1.07)	0.017c (0.64)	0.013b (0.67)	0.002b (0.13)
Transportation accessibility	−0.001 (-0.90)	−0.001 (-1.19)	−0.002 (-2.63)	0.007 (-0.53)
Constant	0.878b (14.48)	2.638a (2.59)	3.062a (1.37)	1.523a (1.82)
City/year fixed effect	YES	YES	YES	YES
N	451	451	451	451
Adjusted R2	0.51	0.68	0.57	0.62

Note.

^ap ≤ 0.01.

^bp ≤ 0.05, and.

^cp ≤ 0.10; t-statistics are in parentheses.

The impact of heterogeneous environmental regulations on IGT is different. Both command and market environmental regulations promote the improvement of GTFP, but voluntary environmental regulation has no significant impact on the improvement of GTFP. In the short term, command and market-based environmental regulations increase the cost of pollution control, and hinder the R&D investment and innovation ability of enterprises. However, in the long term, reasonable ER improve the productivity and resource utilization efficiency of enterprises by the innovation compensation effect, providing evidence for Porter's hypothesis. Compared with command and voluntary ER, market-based ER have a stronger effect on the IGT. However, due to the weak environmental awareness of citizens, it is difficult to have a more direct impact on the production and operation activities of enterprises.

From the perspective of agglomeration economy, the coefficient of industrial agglomeration is positive. This reflects that the agglomeration economic effect promotes the improvement of enterprise productivity and centralized treatment of pollution emissions, thus promoting the IGT. However, the estimated coefficient of resource endowment is significantly non-negative, supporting “resource curse” effect. Auty (2007) firstly put forward the concept of “resource curse”, i.e. abundant resources are a limitation rather than a favorable condition for the economic growth of some countries [35].

4.2. Decomposition of mechanisms

Based on the mediating effect model, the transmission mechanism of MI on IGT is analyzed from the three dimensions (Table 4). Firstly, this study uses the location entropy index to calculate the degree of industrial agglomeration, and analyzes the impact of MI on industrial agglomeration (equations (6), (7))). Secondly, drawing on Cui et al. (2019) [36], the capital and labor mismatch are calculated, and the impact of MI on capital and labor mismatch is analyzed (equations (8), (9))). Again, this study measures the green technology progress using the number of green patents, and based on this, analyzes the impact of MI on green technology progress (equations (10), (11))). Finally, the impacts of industrial agglomeration, capital and labor mismatch, and green technological progress on the GTFP are analyzed separately.

$${Agg}_{it}{=a}_0+{\beta }_1{MI}_{it}+{{{\beta }_{2}SMI+a}_{3}X}_{it}+u_{it}+{\epsilon }_{it}$$ (6)

$${lnY}_{it}{=a}_0+{\beta }_1{MI}_{it}+{{{\beta }_{2}S{MI}_{it}+{{\beta }_{2}Agg}_{it}+a}_{3}X}_{it}+u_{it}+{\epsilon }_{it}$$ (7)

$${Mis}_{it}{=a}_0+{\beta }_1{MI}_{it}+{{{\beta }_{2}S{MI}_{it}+a}_{3}X}_{it}+u_{it}+{\epsilon }_{it}$$ (8)

$${lnY}_{it}{=a}_0+{\beta }_1{MI}_{it}+{{{\beta }_{2}S{MI}_{it}+{{\beta }_{2}Mis}_{it}+a}_{3}X}_{it}+u_{it}+{\epsilon }_{it}$$ (9)

$${Tch}_{it}{=a}_0+{\beta }_1{MI}_{it}+{{{\beta }_{2}S{MI}_{it}+a}_{3}X}_{it}+u_{it}+{\epsilon }_{it}$$ (10)

$${lnY}_{it}{=a}_0+{\beta }_1{MI}_{it}+{{{\beta }_{2}S{MI}_{it}+{{\beta }_{2}Tch}_{it}+a}_{3}X}_{it}+u_{it}+{\epsilon }_{it}$$ (11)

• (1)Agglomeration economy(Agg): There are more indicators and calculation methods for agglomeration economy, and the locational entropy index is used to measure the industrial agglomeration [37].
• (2)Factor mismatch index: This study draws on the practice of Cui et al. (2019) to calculate the capital mismatch index(Mis^k)and the labor mismatch index(is^l)each city separately [36].

$${Mis}_i^k=1\backslash \backslash /Y_i^k-1,{Mis}_i^l=1/Y_i^l-1$$ (12)

In equation (12), $Y_i^k$ denotes the distortion coefficient of regional capital price and $Y_i^l$ denotes the distortion coefficient of regional labor price, which are calculated as follows:

$$Y_i^k=\left(\frac{K_i}{K}/\frac{s_i*{\beta }_i^k}{{\beta }^k}\right),Y_i^l=\left(\frac{L_i}{L}/\frac{s_i*{\beta }_i^l}{{\beta }^l}\right)$$ (13)

In Equation (13), s_i denotes the share of economic output of city i in the total economic output of the city; $\frac{K_i}{K}$ denotes the barrier of the use of capital in city i in the total capital of the city; $\frac{s_i*{\beta }_i^k}{{\beta }^k}$ denotes the share of the use of capital in city i when the capital can be efficiently allocated; ${\beta }_i^k$ denotes the elasticity of the capital output of city when the production function is utilized to calculate the capital output of each city; and $Y_i^k$ denotes the mismatch index of capital in each city. $\frac{L_i}{L}$ denotes the proportion of labor force used by city i to the total labor force; $\frac{s_i*{\beta }_i^l}{{\beta }^l}$ denotes the proportion of labor force used by city i when the labor force can be efficiently allocated; ${\beta }_i^l$ denotes the labor force output elasticity in each city using the production function; and $Y_i^l$ denotes the labor force mismatch index in each city.

• (3)Green technological progress(Tch): Based on the code of the Green Patent List of the International Intellectual Property Organization [38], the number of regional green patents in different years is obtained by searching the official website of the State Intellectual Property Office after setting the patent type IPC code and address of the inventing unit, and used to characterize the level of regional green technological progress.

Table 4.

Mechanisms of MI on the GTFP.

Model	Agg	Misk	Misl	Tch	GTFP
MS	−0.291b (-1.06)	0.457b (2.34)	0.093c (0.41)	−0.764c (-1.98)	−0.054c (-1.90)	−0.062b (-2.14)	−0.057c (-1.93)	−0.055b (-1.93)
SMS	0.203c (1.45)	0.283c (-2.84)	0.025c (0.22)	−0.458b (-2.42)	0.021c (1.41)	0.027c (1.75)	0.023c (1.47)	0.022c (1.42)
Agg					0.015b (0.66)
Misk						−0.012b (-0.59)
Misl							−0.011b (-1.09)
Tch								0.023b (1.76)
Constant	6.997a (18.95)	4.01a(13.31)	3.68a (12.12)	6.022a (17.06)	1.967a (8.83)	1.036a (6.98)	1.041a (7.68)	1.042a (7.22)
Controls	YES	YES	YES	YES	YES	YES	YES	YES
City/year fixed effect	YES	YES	YES	YES	YES	YES	YES	YES
N	451	451	451	451	451	451	451	451
Adjusted R2	0.52	0.47	0.45	0.61	0.51	0.49	0.64	0.59

Note.

^ap ≤ 0.01.

^bp ≤ 0.05, and.

^cp ≤ 0.10; t-statistics are in parentheses.

Firstly, the estimated coefficient of MS on industrial agglomeration is negative, but the coefficient of the SMS is positive, indicating that there is a nonlinear relationship between MS and industrial agglomeration. As the opposite of MS, an increase in the level of MI not only promotes the development of a unified big market, but also reduces inter-regional trade costs and relocation costs, which in turn promotes industrial agglomeration through the local market effect and enterprise relocation effect. However, the coefficient of the quadratic term of MS is negative, reflecting that excessive agglomeration of the manufacturing industry can lead to large pollution emissions and increase the price of production factors, which generates negative externality of agglomeration through crowding effect. In addition, the coefficient of agglomeration economy on GTFP is positive, indicating that MI promotes industrial agglomeration through mechanisms such as sharing, matching and learning, which promotes the improvement of manufacturing productivity and the reduction of resource consumption. However, the coefficient of the quadratic term of MS is negative, reflecting that under the effect of MI, the manufacturing industry in some regions is characterized by over-agglomeration and produces obvious crowding effect, which hinders the IGT.

Secondly, the estimated coefficients of the primary and the quadratic term of MS on capital mismatch and labor mismatch are both significantly positive, indicating that MS significantly exacerbates the mismatch of capital and labor resources, which hinders the improvement of capital and labor utilization efficiency. As the opposite of MS, the development of MI contributes to the free flow of capital and labor factors. Under the action of the competition mechanism, it makes the resources move to regions and industries with high marginal returns. In addition, the estimated coefficient of the capital and labor mismatch on the IGT is significantly negative, reflecting the fact that capital and labor mismatch reduces the resource allocation efficiency and hinders GTFP. Therefore, with the development of MI, the capital and labor's free flow helps to alleviate the distortion of factors and improve their utilization efficiency, which improves industrial productivity and reduces the waste of resources, and promotes the IGT.

Finally, the estimated coefficients of the primary term and the estimated coefficients of the secondary term of MS on green technological progress are both significantly negative, indicating that MS hinders urban green progress and its innovation capacity. As the opposite of MS, MI not only helps to promote the cross-regional flow of innovative factors, but also promotes industrial agglomeration and green technology progress through knowledge spillover effect. In addition, the coefficient of green technology is significantly positive, indicating that the progress of green technology helps to improve the production efficiency and reduce the consumption of resources, which promotes the improvement of the innovation capacity and facilitates its green development.

4.3. Heterogeneity test

With the development of regional integration, the core and edge cities are due to differences in resource endowment and market demand, which lead to industrial structure heterogeneous changes in different regions. This study considers heterogeneity and analyzes the differentiated role of MI on the GTFP in core and peripheral regions. Specifically, drawing on the methodology [39], 16 cities before the expansion of the YRD before 2010 are regarded as the core cities, and 25 cities joined after 2010 are regarded as the peripheral cities (Table 5).

Table 5.

Impact of MI on the GTFP in different regions.

Variable	Core cities	peripheral cities
MS	−0.128b (-2.28)	−0.102 (-1.97)
SMS	0.059c (1.94)	−0.062b (-2.19)
GTFPt-1	0.648a (10.63)	0.57a (8.11)
Command environmental regulation	−0.001c (-0.76)	0.002c (1.37)
Market environmental regulation	0.013a (4.01)	0.025a (6.98)
Voluntary environmental regulation	−0.135 (-1.34)	0.142 (1.28)
Agglomeration economy	0.019 (1.51)	0.015b (2.49)
Resource endowment	−0.007 (-1.33)	0.002b (1.99)
Industry structure	0.154b (2.610	0.017 (0.40)
Energy consumption structure	−0.817a (-6.26)	0.151 (2.28)
GDP	0.032a (2.97)	−0.003 (-0.25)
FDI	0.014b (2.19)	0.008a (2.67)
Urban innovation	0.01a (1.43)	−0.008b (-1.56)
Urbanization rate	0.046 (1.52)	0.038 (1.37)
City size	0.033 (0.87)	0.037 (0.97)
Transportation accessibility	−0.001c (-1.68)	0.001 (0.37)
Constant	1.119a (9.97)	0.598a (5.66)
City/year fixed effect	YES	YES
N	176	275
Adjusted R2	0.47	0.63

Note.

^ap ≤ 0.01.

^bp ≤ 0.05, and.

^cp ≤ 0.10; t-statistics are in parentheses.

The results show that the MS has a significant negative impact on the IGT in the cities of origin, but not in the newly joined cities. Since MI is the opposite of MS, this result indicates that MI promotes the GTFP in core cities, but has a non-significant impact on the IGT in newcomer cities. The core cities joined earlier and have a relatively high level of MI, and after years of development, they have initially entered the stage where the center city radiates and drives the development of the surrounding areas. As a result, the core cities have the advantages of information, technology and capital to promote the regional manufacturing production efficiency. In addition, the economic development level of core cities is generally relatively good. With the improvement of MI level, the core city can concentrate advantageous resources to develop high-end technology manufacturing industry and purchase low-end industrial products through trade, so as to reduce the pollution emission and promote the optimization of industrial structure through trade instead of production. However, the marginal cities generally lag behind in economic development, and even become the key transfer areas for pollution-intensive industries in the core cities. In particular, newcomers such as Huaian in northern Jiangsu and Chizhoug in Anhui have taken over more pollution-intensive industries from their cities of origin, resulting in a failure of MI to significantly promote the IGT in the peripheral cities of the YRD.

The quadratic coefficient indicates that MS has a positive effect on the IGT in the core cities, but a significant negative effect on the IGT in the edge cities. With the continuous development of MI, the core city attracts more manufacturing enterprises by virtue of its good economic development level, infrastructure and innovation environment, which improves the productivity of enterprises and promotes the IGT BY the agglomeration economic effect. However, with the massive concentration of manufacturing enterprises, production factors such as land and labor continue to rise, resulting in serious environmental pollution, thus generating negative agglomeration externalities and impeding the IGT. Therefore, with the continuous development of MI, although the core cities improves the IGT through industrial transfer and industrial agglomeration, they also need to pay attention to the negative externality of agglomeration and its resultant environmental pollution in the manufacturing industry. Although the peripheral cities take over the pollution-intensive industries of the core cities and hinder the GTFP, with the continuous improvement of MI, the accelerated flow of innovative factors and the knowledge spillover effect will lead to the improvement of the innovation ability of enterprises in the new cities and the improvement of pollution control technology, which will promote the IGT.

4.4. Robustness test

This study utilizes the replacement of core variables and regression methods for robustness testing (Table 6). Firstly, calculate the pollution emission index of the industry and utilize the dynamic panel to analyze the impact of MI on the pollution emission and measure the IGT from the perspective of pollution emission. Secondly, calculate the share of output value of pollution-intensive industries and analyze the impact of MI on the share of pollution output value of industry by dynamic panel method to measure the IGT from the structural perspective. Finally, to solve the endogeneity problem, the 2SLS method is used to analyze the impact of MI on the pollution emission of industry and the share of output value of polluting industry, respectively.

Table 6.

Robustness test of MI affecting GTFP.

Model	Pollution emission		Pollution industry output
Variable	DPD	2SLS	DPD	2SLS
MS	0.016b (0.44)	0.073c (-1.12)	0.169c (0.19)	0.073c (1.12)
SMS	−0.009b (-0.46)	−0.035b (1.08)	−0.102c (-0.20)	−0.035b (-1.08)
GTFPt-1	0.482a (9.74)	0.161b (5.32)	0.292b (7.25)	0.309a (8.47)
Command environmental regulation	−0.002b (-1.01)	−0.005a (-4.30)	−0.021c (-0.74)	−0.005b (-4.30)
Market environmental regulation	−0.033a (-3.44)	0.007b (2.07)	−0.019b (-0.18)	−0.007a (-2.07)
Voluntary environmental regulation	0.043 (0.66)	0.159 (1.56)	0.03 (0.02)	0.159 (1.56)
Agglomeration economy	0.004a (0.24)	0.018a (2.77)	0.279b (1.28)	0.018b (2.77)
Resource endowment	−0.009b (-1.40)	0.006 (5.31)	−0.145c (-1.16)	−0.006a (-5.31)
Industry structure	0.199b (2.41)	−0.155a (-4.35)	−0.701a (-0.54)	−0.155a (-4.35)
Energy consumption structure	0.045a (0.38)	0.482a (6.34)	1.462b (0.76)	0.482a (6.34)
GDP	0.063b (2.09)	0.036a (4.29)	0.136 (0.53)	0.036a (4.29)
FDI	−0.002c (-0.49)	0.021a (5.56)	0.067b (0.54)	0.021a (5.56)
Urban innovation	−0.004b (-0.36)	0.005b (0.80)	0.074b (0.64)	0.005c (0.80)
Urbanization rate	0.056c (1.63)	0.033c (1.29)	0.446c (0.87)	0.033c (1.29)
City size	0.013b (0.27)	−0.004 (-0.13)	−0.438 (-0.61)	−0.004 (-0.13)
Transportation accessibility	−0.001 (-1.43)	0.001(0.96)	−0.052 (-2.31)	0.001 (0.96)
Constant	−1.72a (-6.16)	−6.12a (-16.19)	2.405a (0.78)	−0.61a (-6.19)
City/year fixed effect	YES	YES	YES	YES
N	451	451	451	451
Adjusted R2	0.61	0.69	0.41	0.52

Note.

^ap ≤ 0.01.

^bp ≤ 0.05, and.

^cp ≤ 0.10; t-statistics are in parentheses.

From the perspective of pollution emission, the coefficient of MS is positive, but the SMS is negative. This indicates that there is also a nonlinear relationship between MS and pollution emission. The MS is controlled within a certain range, which is favorable to the reduction of manufacturing industry's pollution emission and promotes the IGT. Meanwhile, after regression by 2SLS, the estimated coefficient of MS is positive, and its quadratic term coefficient is negative. Since MI is the opposite of MS, the empirical results show that with the increase of MI, the pollution emission also shows a tendency of decreasing and then increasing, the MI promotes the reduction of pollution emission and its green transformation, which suggests that the conclusions of the previous study have a good degree of robustness.

From the structural perspective, the estimated coefficient of MS is positive, and the estimated coefficient of SMS is negative, indicating that there is a nonlinear relationship between MS and pollution industrial output, and that MS controls a certain range to help pollution-intensive output decline, thus promoting the IGT. In addition, after replacing 2SLS for regression, the coefficient of the MS is positive, while the coefficient of the secondary term is negative. As the opposite of MS, the empirical results also show that there is a non-linear relationship between MI and the proportion of pollution-intensive industry output value. When the MI level is low, the cost of inter-regional commodity trade, the cost of factor mobility, and the cost of enterprise relocation are high, which leads to the development of local governments to mainly develop the industry of their own jurisdictions to promote economic growth, and even at the expense of the introduction of a large number of pollution-intensive industries, which hinders the IGT. As the level of MI increases, enterprises move more freely and the allocation efficiency of production factors increases, which promotes the transfer of pollution-intensive industries across regions or sectors and affects the GTFP. Therefore, from a structural perspective, there is also an inverted U-shaped relationship between MI and IGT, which improves the robustness of the previous estimation results.

4.5. Spatial effect analysis

Due to the possible spatial correlation of non-desired outputs such as pollution emissions, analyzing the impact of MI on the IGT still needs to take into account the effect of neighboring regions. Therefore, this study measures the Moran I index of manufacturing GTFP and utilizes the spatial lag model and the spatial error model to further examine the impact of MI on GTFP. In the spatial measurement regression, the correlation coefficient of the spatial lag term of the GTFP is positive, reflecting that there is spatial correlation in the IGT, and it is affected by the level of IGT in neighboring regions (Table 7). In addition, the estimated coefficient of the spatial lag term of MS is significant. This indicates that the MS in neighboring regions hinders the IGT, and the development of the level of MI in the neighboring regions promotes the IGT. As the level of MI in neighboring regions increases, local goods enter neighboring regions more freely. This not only helps to improve inter-regional resource allocation efficiency, but also promotes innovative factor flow. This further promotes the GTFP of inter-regional industries. In addition, there is also a positive correlation between the MI of neighboring regions and the level of local MI, the improvement of MI in neighboring regions also helps to promote the improvement of local MI and IGT.

Table 7.

Spatial econometric regression of MI on GTFP.

	Geographic binary matrix		Geographic distance matrix
Model	SLM	SEM	SLM	SEM
WcGTFP	0.011c (2.24)		0.059c (4.11)
WcMS		−0.481c (-1.861)		−0.050c (-0.250)
MS	−0.083b (-0.484)	−0.124b (-5.16)	−0.094b (0.551)	−0.094b (0.5520
SMS	0.044c (4.83)	0.067c (2.71)	0.049c (5.37)	0.051c (5.51)
Command ER	0.001 (0.520)	−0.001* (-0.432)	0.002c (0.616)	0.001 (0.466)
Market ER	0.019b (2.3980	0.025a (3.062)	0.019b (2.348)	0.019b (2.398)
Voluntary ER	0.987 (3.389)	1.007 (3.267)	0.188 (3.408)	0.085c (3.387)
Constant	0.714a (2.212)	0.702a (2.471)	0.653 (2.042)	0.697 (2.283)
Controls	YES	YES	YES	YES
City/year fixed effect	YES	YES	YES	YES
N	451	451	451	451
Adjusted R2	0.62	0.64	0.53	0.58

Note.

^ap ≤ 0.01.

^bp ≤ 0.05, and.

^cp ≤ 0.10; t-statistics are in parentheses.

5. Conclusion

As an important dimension of regional integration, MI has a profound impact on socio-economic development. Based on urban panel data, this paper analyzes the impact of MI on the IGT by dynamic panel GMM model and spatial econometric model. The main conclusions are as follows:

• (1)The MI has a nonlinear relationship on IGT, and with the increase of the level of MI, IGT shows a tendency of rising first and then declining. In addition, different environmental regulations have differentiated effects on IGT. And while the command and market types have significantly promoted IGT, the impact of voluntary environmental regulations is low.
• (2)The MI has a significant impact on IGT mainly through industrial agglomeration, resource mismatch and green technological progress. MI promotes industrial agglomeration and IGT through local market effect and enterprise migration effect. However, excessive industrial agglomeration can also lead to congestion effects and hinder the IGT. At the same time, MI improves the free flow of factors of production, and reduces the mismatch of capital and labor which improved the efficiency of factory utilization. What's more, MI also promotes inter-regional knowledge spillovers and its green technological advances, and promotes IGT.
• (3)The effect of MI on IGT exhibits spatial heterogeneity. Although MI significantly promotes the IGT in core cities, the impact on the IGT in marginal cities is not significant. By replacing the core indicators and regression methods, the impact of MI on IGT remains robust.
• (4)There is a certain spatial spillover effect of IGT, the MS in neighboring regions will hinder the IGT in the region. As the MI of neighboring regions continues to improve, it helps to improve the inter-regional resource allocation efficiency, and promotes the innovation factors flow, thus promoting the IGT.

This study has important policy implications for the construction of a unified big market and industrial transformation and upgrading. Firstly, break local protectionism and eliminate inter-regional trade barriers, and accelerate the construction of regional and national unified big market. Secondly, increase the reform of the factor market and promote the free flow of factors across regions, so as to improve the efficiency of the allocation of factors of production and promote industrial transformation and upgrading. It is necessary to take into account the level of economic development of different regions in order to implement differentiated strategies. For regions with higher levels of economic development, the government should intervene moderately in the allocation of resources and the assumption of responsibility for pollution control, so as to reduce the over-concentration of resources in the core city and its negative externality of agglomeration. For regions with lower levels of economic development, environmental regulations should be moderately increased to avoid becoming pollution haven. Finally, local governments need to strengthen collaborative development, harmonize the implementation standards of inter-regional environmental regulations, and carry out joint pollution prevention and control, so as to promote the overall pollution reduction and green transformation of industries.

Compared with previous studies, this study measures the IGT through the super-efficient SBM model, and analyzes the impact of MI on IGT and its mechanism by the dynamic panel GMM model. This not only enriches the understanding of the resource-environmental effect of MI, but also has important policy significance for promoting the integrated development of urban agglomeration. However, this study also has certain limitations and points out directions for future research. Due to data limitations, this paper mainly analyzes the impact of MI on the GTFP of urban industry, but it lacks the ability to analyze the role of MI on IGT at other scales. Future research can also analyze the resource-environmental effects of MI at different scales.

6. Ethics declarations

• •Review and/or approval by an ethics committee was not needed for this study because this is an observational social science study that does not involve experiments with humans or animals.
• •Informed consent was not required for this study because the research does not involve experiments on humans or animals, and the data used has been downloaded and utilized in accordance with relevant regulations from the respective official websites.

Funding

This study was funded by the Ningbo Philosophy and Social Science Planning Project “The theoretical mechanism and practical path of Ningbo's digital industry agglomeration driving green and low-carbon development under the background of double cycle” (G2024-1-21), "National Tax System Tax Research and Research Base" jointly built by Ningbo Yinzhou District Taxation Bureau and Zhejiang Industrial and Commercial Vocational and Technical College, and "Ningbo Jinsui Tax Big Data Research Base", the sixth round of Ningbo Philosophy and Social sciences research base.

Availability of data and materials

Data was collected from the China City Statistical Yearbook. Data is freely available at: https://www.epsnet.com.cn/index.html#/Index.

CRediT authorship contribution statement

Chenghui Tang: Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.