Research on the Mechanism of Transfer Payment Policy on Resource Dependence of Resource-Depleted Cities

Abstract

The objective of this study is to clarify the impact and mechanism of transfer payment policies on the resource dependence of resource-depleted cities. Based on the panel data of 113 prefecture-level resource-based cities from 2006 to 2017, this study uses a multi-period difference-in-differences model to conduct an empirical study on the impact and mechanism of transfer payment policies on resource-depleted cities. The results are as follows: Firstly, the transfer payment policy can reduce the resource dependence of resource-depleted cities. Secondly, there is a significant difference between the eastern region and the central and western regions in terms of the effects of policy implementation. Thirdly, transfer payment policies reduce local dependence on resources, mainly through upgrading industrial structures and enhancing infrastructure construction and technological progress. The research indicates that providing financial policy support for the transformation of resource-depleted cities, exploring ways to reduce resource dependence in the eastern region, playing an exemplary role, and expanding the intensity of urban industrial transformation are of reference significance for the sustainable development of resource-depleted cities. This study also contributes to the coordinated development of the regional economy and the policy formulation of the sustainable development of resource-depleted cities.

1. Introduction

Resource-based cities are cities whose development mainly depends on the exploitation and processing of non-renewable resources such as minerals, forests, and petroleum in the region. There are 262 resource-based cities in China, accounting for 40% of the total number of cities in China. Since 1949, a total of 5.8 billion tons of iron ore, 52.9 billion tons of raw coal, and 5.5 billion tons of crude oil have been mined, and this has made an indelible contribution to the economic development of resource-based cities. However, as the non-renewable resources in these cities began to be exhausted, some of these cities have less than 30% recoverable reserves of natural resources and become resource-depleted cities. Over-reliance on resource-based industries crowded out other local industries and this phenomenon has been known as the “Dutch Disease” or “Resource Curse” [1,2]. Shao and Qi have proved that there exists a serious “Resource Curse” phenomenon in China [3]. James and Aadland pointed out that enterprises with high pollution, high emissions, and low efficiency lead to the emission of a large amount of carbon dioxide and air pollutants, causing serious damage to the ecological environment [4]. Ross showed that over-dependence on natural resources makes resource-based cities fall into the economic development dilemma of a single industrial structure, a high proportion of “three high” industries, and a low innovation capability [5]. During the process of continuous advancement of energy conservation and emission reduction in China, resource-depleted cities not only face the pressure of seeking new economic growth but also face the enormous pressure of energy conservation and emission reduction. The economic and social contradictions of resource-depleted cities are the concentrated embodiment of the problems faced by high-quality development. Therefore, it can be seen that whether resource-depleted cities can reduce their dependence on resources and realize transformation and upgrading is the key to their high-quality economic development, which is of great significance for local social stability and economic development.

How to reduce the resource dependence of resource-depleted cities, help them out of the “Resource Curse” dilemma, and complete the economic transformation have always been the focus of scholars. Resource-based cities can improve the efficiency of transformation through economic and social environment construction, the inflow of new industrial factors, the improvement of existing resource allocation levels, and the improvement of allocation efficiency [6]. Political incentives and performance competition affect the transformation efficiency of resource-depleted cities, especially the solution to environmental pollution [7,8]. At the same time, most studies on the transformation of resource-depleted cities take specific cities as case studies. For example, Wen et al. take Benxi as an example to prove that the transformation of the original industry and the elongation of the industrial chain through high-tech and modern low-carbon and environmental protection can contribute to the transformation process [9]. Research based on the carbon emission data of Xuzhou in Jiangsu province showed that there is a decoupling between the urbanization transition and the environment of resource-depleted cities [10]. A study about the ecological transformation of six resource-depleted cities in Jilin province found that the level of science and technology can promote ecological efficiency [11]. In recent years, with the promulgation of the <National Sustainable Development Plan for Resource-Based Cities (2013–2020)> (hereinafter referred to as the <Plan>), the role of policies in the transformation of resource-based cities has gradually attracted scholars’ attention. The study evaluated the environmental impact of the <Plan> on resource-based cities and demonstrated through experiments that the introduction of the <Plan> can help reduce the emission intensity of pollutants in resource-based cities [12]. The supportive policies for resource-depleted cities could significantly improve per capita GDP and employment rate [13]. It was proved that transformation policies can promote the economic growth of resource-depleted cities by constructing a heterogeneous time-series difference-in-differences model [14]. By constructing the fixed-effect model and the dynamic system generalized distance model (SYS-GMM), Zhang et al. concluded that the governance ability of resource-based cities has an important influence on the degree of urban resource dependence and proposed different measures in, and effects on, the eastern region, northeast region, and declining cities [15].

The Chinese government has been making continuous efforts to reduce the dependence of resource-depleted cities on resources and promote the transformation and upgrading process of resource-depleted cities. In order to advance the process, the State Council promulgated the <Several Opinions of the State Council on Promoting the Sustainable Development of Resource-depleted cities> in 2007, proposing a policy of direct financial transfer payments from the central government to resource-depleted cities. Subsequently, the National Development and Reform Commission, the Ministry of Land and Resources, and the Ministry of Finance included 69 resource-depleted cities (including county-level cities and municipal districts) in the support list of financial transfer in three batches in 2008, 2009, and 2012 and arranged for the central government to provide annual subsidies to resource-depleted cities, with a total of nearly 160 billion yuan in financial transfer payments. The goal is to help resource- depleted cities establish and improve the compensation mechanism for resource development and the assistance mechanism for declining industries so that they can improve the efficiency of resource utilization, reduce the degree of resource dependence, complete the transformation and innovation, and realize the green and sustainable development of the urban economy.

Existing research has made a lot of evaluations of transfer payment policy in economic development, green innovation, and environmental protection in resource-depleted cities. Supporters believed that this policy encouraged enterprises to make capital investments by providing capital subsidies, tax cuts, and fee reductions, so as to promote technological progress and industrial upgrading of production and achieve higher value-added production activities [16,17]. Xu and Tan affirmed the role of transfer payments in increasing per capita GDP in resource-depleted cities [18]. A study found that a transfer payment policy can promote green technology innovation in resource-depleted cities, and the policy effect is strengthened over time [19]. Based on the enterprise data, the transfer payment policy could promote employment and improve output [18]. Based on the difference-in-differences model, the results confirmed the effect of the transfer payment policy on urban carbon emission reduction [20]. Critics have different opinions as follows. Due to the information asymmetry between the central government and local government, the transfer payment policy would induce the local government to reduce tax collection and reduce the level of fiscal effort, so as to obtain more benefits from transfer payment [21,22]. López-Laborda and Julio pointed out that the “common pool problem” of transfer payments led to the phenomenon of “fiscal illusion” in local governments [23]. The benefits of local public services financed by transfer payments were enjoyed by regions, while most of the costs were borne by other regions, which ultimately led to deviations in the incentive mechanism for local government behavior. Liu pointed out that more transfer payments would not necessarily lead to faster economic growth, and transfer payments would have a negative impact on the economic growth of backward regions without a good system [24]. They believed that the transfer payment policy would make resource-depleted cities fall into an “incentive trap”, which led to slower economic growth [16]. The government was short-sighted in the use of transfer payment, investing, and supporting large state-owned enterprises in order to ensure regional GDP growth and employment, but these enterprises were usually highly polluting and inefficient [25]. Increased government intervention had a significant inhibitory effect on local innovative development [26]. The local area was caught in a vicious circle in which green technology innovation was not valued and the natural environment was continuously damaged.

At present, there is no specific literature on the impact of financial transfer payment policy on resource endowment, and the corresponding mechanism is not clear. A few studies only focus on specific urban cases and lack macro-analysis at the overall level. In order to explore the correlation between transfer payments and resource dependence, this study uses the annual data of 113 resource-based cities in China from 2006 to 2017 and uses the difference-in-differences model to conduct a purely empirical study basically on the impact degree and mechanism of transfer payments and resource dependence. The data are from the China Statistical Yearbook 2006–2017. This study also explains the mechanism of transfer payment affecting resource dependence through industrial transformation and explains the different effects of policies in different regions through heterogeneity analysis. Because the financial transfer payment policy only affects resource-depleted cities, it provides a natural condition for the difference-in-differences model to divide the experimental group and the control group. Meanwhile, the transfer payment policy is carried out in three batches, and the effect time of each policy is more than 3 years. The data structure is suitable for using the multi-period difference-in-differences model. Compared with the general regression method, the biggest advantage of the difference-in-differences model is that it can avoid the endogeneity problem. Since policies are generally exogenously compared with microeconomic subjects, there is no reverse causality problem. The fixed-effect estimation in the difference-in-differences model can also alleviate the missing variable bias problem. Considering the advantages of the difference-in-differences model, we choose it as the experimental method in this paper.

From various perspectives, this study provides the following three theoretical contributions. Firstly, the research explores the effect of the transfer payment policy on resource dependence of resource-depleted cities at the level of prefecture-level cities, which fills the gap in the research of related fields and contributes to the sustainable development goals of resource-based cities in China. Secondly, this study discusses the ways in which transfer payments affect local resource dependence and selects control variables to measure regional heterogeneity, including economic development level parameters (lngdp) and (lngdpp) [27,28], government size parameter (lngov) [29], urbanization level parameter (lnden) [30], industrial development scale parameter (lnqys) [31], urban infrastructure level parameter (lnpas) [32], urban pollutant emission level parameter (lnwat) [33], urban pollutant treatment level parameter (su) [34], science and technology investment level parameter (sci) [35] and the education investment level parameter (edu). This study quantitatively analyzes the function of industrial transformation and puts forward some suggestions for the implementation of transfer payment or similar policies. Finally, the research conducts a spatial heterogeneity analysis of the effect of the transfer payment policy on resource dependence in different regions of China. Based on the results, we recommend the Chinese government formulate policies according to local conditions to ensure that resource-depleted cities in various regions achieve the goal of sustainable development as soon as possible.

2. Research Hypothesis

Financial transfer payment policies mainly include general transfer payments, transfer payments for ethnic minority areas, government awards and subsidies for counties and townships, transfer payments for wage adjustments, transfer payments for rural tax and fee reform, and financial subsidies for year-end settlement. The financial transfer payment policy undertakes the important mission of establishing and improving the long-term sustainable development mechanism of resource-depleted cities, cultivating replacement industries, solving social problems such as employment, and strengthening environmental governance and ecological protection. It is one of the important measures taken by the state to promote the sustainable development of resource-based cities. In the context of a freer flow of factors and a more open economy, unbalanced and inadequate financial, resource, and environmental problems in regional economic development have become a key link that restricts China’s high-quality economic development. As an important part of regional economic development, the development of resource-depleted cities needs to shift from the traditional factor-driven growth model to a new growth model focusing on improving quality and efficiency urgently [36]. The research results of foreign scholars on regional economic development show that the coordination relationship between regions can be comprehensively considered from the perspective of equilibrium analysis framework, spatial perspective, and government intervention [37,38,39,40,41,42,43,44]. Chinese scholars have also understood the interaction mechanism between increasing returns, knowledge, human capital, innovation, regional disparities [45,46,47,48], etc., from the perspective of regional science and new economic growth theory. This provides a theoretical explanation and policy foundation for the sustainable development of resource-depleted cities to promote regional coordinated development in China. Resource-depleted cities play an important role in China’s sustained regional economic growth and ensuring the supply of resources and energy, which is an important link to the high-quality and high-level development of China’s economy, and financial transfer payment is an important policy means for the government to regulate regional economy [49]. According to the transmission path of the transfer payment effect, the impact of the transfer payment on regional economic development is transmitted through the local government fiscal policy. Whether transfer payment can reduce the degree of resource dependence of resource-depleted cities is critical to whether it can promote the sustainable economic development of these cities and thus regional economic sustainable development. Therefore, this study focuses on the impact of the transfer payment policy on the resource dependence of resource-depleted cities and its mechanism, and the research hypothesis is put forward accordingly.

According to the existing research, economic logic, and historical experience, transfer payments for resource-depleted cities are mainly used to solve the problems in non-resource industry continuation, infrastructure construction, environmental protection, and green technology innovation of resource-depleted cities [19]. Firstly, the finance transfer payment policy should play a guiding role in the adjustment and industrial structure transformation of resource-depleted cities, the vigorous development of tertiary industry, accelerating the replacement and replacement of non-resource-based enterprises, and encouragement of innovative development of small and medium-sized enterprises by optimizing the structure of local budget expenditure, so as to provide a new perspective and ideas for regional industrial transformation. Ma and Yu point out that support policies for resource-depleted cities can promote the transformation and upgrading of the manufacturing industry through the effect of resource reallocation [50]. Secondly, policy needs to be supportive. Transfer payment policies can effectively alleviate the plight of the local government’s fiscal deficit and help the government to increase the proportion of scientific and technological expenditure and education expenditure to improve local infrastructure construction, such as the medical and health system and education system [51]. Thus, it is conducive to the introduction and cultivation of local high-tech talents. Finally, the policy has to play an incentive role. By reducing taxes and fees and increasing subsidies, high pollution and high emissions enterprises have more motivation and willingness to innovate green technologies and improve their production efficiency [3]. This can achieve the purpose of reducing the discharge of pollutants, protecting the local environment, enabling resource-depleted cities to enter into healthy development, and reducing resource dependence. Accordingly, this study proposes the following research hypotheses:

H1. 

Transfer payment policy helps reduce the resource dependence of resource-depleted cities.

Furthermore, referring to the polarization theory of regional economic development, economic development has a tendency towards nonequilibrium and divergent evolution. The non-homogeneity and illiquidity of production factors in economic development lead to the inability to replace production factors that hinder balanced development in various regions, making the process of economic development not leading to equilibrium but to the strengthening of regional differences [24]. On the one hand, the non-homogeneity and immobility of the production factors in economic development lead to the fact that production factors cannot be completely replaced in various places, so the process of economic development does not lead to equilibrium but strengthens regional differences. On the other hand, markets are not characterized by perfect competition, but by monopoly, oligopoly, and externalities. Important information about technology and innovation is not freely available but needs to be disseminated through the economic system. Myrdal’s cyclic accumulation process theory emphasizes the polarization effect [39]. Active local development will inhibit the development of its surrounding areas. For example, prosperous areas attract high-skilled labor forces in stagnant areas, which weakens the innovation potential of surrounding areas, damages the surrounding environment, and intensifies the competition of production factors, making stagnant areas fall into a vicious cycle. Therefore, the stimulus will accumulate over time and form a fixed gap, and the non-equilibrium state will be reinforced with economic development.

There are huge differences in endowments (such as infrastructure, economic development, social development, and marketization) among the eastern, central, and western regions of China, and such differences continue to expand with the flow of capital and labor [52,53].

Due to its proximity to trading ports, the resource-based cities in the eastern region are superior to those in the central and western regions in terms of economic aggregate, scientific and technological innovation, talent training and attraction, and the overall development of industrial structure is relatively complete [54]. In terms of environmental protection issues, the eastern region has stronger environmental awareness and higher environmental protection standards. Meanwhile, many high-polluting industries have been relocated to the underdeveloped western region. Most of the cities in the central region are dominated by industry and agriculture, and the service industry started late. They also have many problems in economic development, such as an unreasonable development of resources and environmental pollution. The economic development in western China is rough, with backward technology and serious environmental pollution [55]. Therefore, if the transfer payment policy is implemented for resource-depleted cities in the three regions, the effects of the policy may be quite different. One view holds that resource-depleted cities in the eastern region have a better policy environment, so they are more motivated to reduce resource dependence than cities in the central and western regions. Another point of view is that the mature development of the eastern region will reduce the marginal effect of policy, but the resource-depleted cities in the central and western regions can reduce resource dependence more effectively with the support of policies. Based on this, the study proposes the second hypothesis:

H2. 

There is regional heterogeneity in the effects of the transfer payment policy. The eastern, central, and western regions will have different responses to the transfer payment policy.

Lastly, industrial structure transformation has always been the key factor in promoting urban transformation. Resource-based cities take advantage of their own resource endowment to gather resource-intensive industries and their supporting industries, forming a “heavy” industrial structure [56]. Taking coal resource-based cities as an example, their industries start from coal and gradually form an industrial chain, including coal mining, processing, sales, and other links under the dual influence of the scale and agglomeration effect. Their industrial structure is deeply influenced by coal resource endowment, making economic development constrained by resource endowment and falling into the “dilemma” of the original industrial structure [57]. Resource-based cities rely on local resources and concentrate on the exploitation and rough processing of core resources, which will have a crowding out effect on high-quality human capital and technical elements, solidifying the original industrial structure [58], which is not conducive to the upgrading of the industrial structure of resource-based cities.

Industrial structure transformation and upgrading can be achieved through the improvement of industrial structures and rationalization of industrial structures [31]. The improvement of industrial structures is the process of the development of industrial structures from a low level to a high level. In a narrow sense, it refers to the process of industrial structures changing from labor-intensive to capital-intensive structures and then to technology-intensive structures. In a broad sense, it refers to the process of industrial structures changing from primary industries to secondary and tertiary industries. The rationalization of industrial structures represents the enhancement of the coordination ability and the improvement of correlation between different industries, which reflects the aggregation quality among industries. When the evolution of industrial structures synergistically drives the improvement of labor productivity within each industry and the industry with higher labor productivity accounts for a larger share, it belongs to the advanced industrial structures of high quality. Appropriate policies can lead to the innovation and development of enterprises and help enterprises improve their innovation efficiency [22]. As the subsidies are directly paid by the central government to the local government, the transfer payment policy can assist the local government to provide financial support and tax incentives for the technological innovation of enterprises, lead the production factors to flow from low-efficiency enterprises into high-efficiency enterprises, and drive the industrial structures from a low level to a high level. Finally, the transformation of local industrial structures will be promoted, tertiary industries will be the focus of economic development, and the mining of non-renewable resources will be reduced.

The industry needs correct guidance in the process of development to avoid resource misallocation and resource waste caused by blind investments and overproduction. The goal of industrial structure rationalization is to help enterprises reduce the friction on the unreasonable industrial structure, reduce the replacement cost of the enterprises, make up for the incomplete market information, and improve the efficiency of resource allocation among industries. At the same time, strengthening infrastructure construction is conducive to cultivating local technical personnel and attracting more high-quality talents [54]. So that strengthening infrastructure and technological progress are also important factors driving urban transformation. Wherein, infrastructure construction includes a series of measures such as local investment in science and technology, improvement of the education system and medical system, construction of urban roads, and urban greening. Technological progress represents the goal of protecting the natural environment and ecology and achieving sustainable development by means of green technology innovation, improving production efficiency, and reducing pollutant emissions. Based on the above analysis, the third hypothesis is proposed:

H3. 

The transfer payment policy affects the resource dependence of resource-depleted cities through industrial transformation, infrastructure construction, and technological progress.

3. Methodology

3.1. Model Setting

After the State Council formulated and promulgated <Several Opinions of the State Council on Promoting the Sustainable Development of Resource-Based Cities> on 24 December 2007, China successively identified national resource-depleted cities in three batches in 2008, 2009, and 2012. There are 12 cities in the first batch, 32 cities in the second batch, and 25 cities in the third batch, making a total of 69 resource-depleted cities, including prefecture-level, county-level, and municipal districts. This policy constructs an experiment for the study of using the difference-in-differences model to assess the impact of transfer payments on regional resource dependence. Considering that the development and resource abundance of different cities in China is quite divergent, this study controls the research scope to resource-based cities identified in <National Sustainable Development Plan for Resource-Based Cities (2013–2020)>. In view of the availability of data, a total of 113 resource-based cities at the prefecture-level city scale are selected as research samples, 24 of which are resource-depleted cities as the experimental group, and the remaining 89 cities as the control group. On this basis, referring to the practice of Song, Liu, and Zhao [19,59], the corresponding multi-period difference-in-differences model is constructed as follows:

$$r{d}_{it}={\beta }_0+{\beta }_{1}DI{D}_{it}+{\beta }_{2}contro{l}_{it}+{\eta }_i+{\gamma }_t+{\epsilon }_{it}$$

(1)

In Formula (1), subscripts $i$ and $t$ represent the region and year, respectively. The explained variable $r{d}_{it}$ is the resource dependence degree of city $i$ in year $t$. $DI{D}_{it}$ is the core explanatory variable, namely, the multi-period difference-in-differences variable ($DI{D}_{it}=treatmen{t}_i\times pos{t}_{it})$. $treatmen{t}_i$ represents whether it is a resource-depleted city (if so, then $treatmen{t}_i=1$) (otherwise, $treatmen{t}_i=0).$ $pos{t}_{it}$ represents whether the policy is implemented in city $i$ in year $t$ (if so, then $pos{t}_{it}=1)$ (otherwise, $pos{t}_{it}=0)$. $contro{l}_{it}$ represents the control variables that affect the degree of resource dependence with respect to $i$ and $t$. ${\eta }_i$ represents the individual fixed effect, which controls the individual factors that affect the degree of resource dependence but does not change with time. ${\gamma }_t$ represents the time effect, which controls the time factor affecting all regions with the time change, and ${\epsilon }_{it}$ is the error term. The positive and negative signs and numeric values of ${\beta }_1$ reflect the direction and degree of influence of the transfer payment policy in resource-depleted cities on local resource dependence.

3.2. Description of Variable and Data

3.2.1. Explained Variable and Core Explanatory Variable

The explained variable is regional resources dependence (rd). Resource dependence refers to the amounts of natural resources required for regional economic development, and it is usually a ratio indicator (such as the proportion of resource industry output value in GDP [60] and the proportion of employees in the resource industry) that accounts for the total number of employees [61]. This study refers to the ratio of the mining industry employees to the total population at the end of the year [62] to measure the resource dependence of the region. Because the mining industry covers a wide range, including coal, oil, natural gas, metal and non-metallic mineral processing, wood harvesting and other industries directly related to natural resources, it can comprehensively measure the economy’s dependence on natural resources. At the same time, the total population at the end of the year is a more stable statistical indicator than the employment population at the end of the year, which can make the data more robust.

DID is the core explanatory variable, representing whether the locality is classified as a resource-depleted city and provided with corresponding transfer payments.

3.2.2. Control Variables

Since the degree of resource dependence is closely related to the regional economic and social development, in order to make the degree of resource dependence of the experimental group and the control group comparable, it is necessary to control the characteristics that affect the economic and social development of the experimental group and the control group. The control variables selected in this study to measure regional characteristics include (1) the level of economic development, specifically the logarithm of real GDP per capita (lngdp) and its quadratic term (lngdpp), of which the GDP value is deflated with 2006 as the base year. Per capita, GDP can directly reflect regional economic conditions and economic development levels [27]. Meanwhile, the quadratic term of per capita GDP is often used to test the classical Environmental Kuznets Curve (EKC) and analyze the “inverted U-shaped” relationship between economic development level and environmental indicators [28], so as to control the influence of the economic development degree on the resource dependence degree. (2) The government’s size, specifically the local public budget expenditure (lngov), will affect the degree of environmental emphasis of local governments in the process of economic development. Reasonable fiscal expenditure can improve the level of public services and help reduce resource dependence [29]. (3) The level of urbanization, specifically the logarithm of population density (lnden), reflects the level of urbanization to some extent. The more concentrated the population density, the higher the level of economic development of a city, thus influencing the degree of resource dependence [30]. (4) The industrial development scale is the number of industrial enterprises above the designated size (lnqys). The more industrial enterprises there are, the heavier the local dependence on the secondary industry will be and the higher the dependence on non-renewable resources will be [31]. (5) The level of urban infrastructure is the logarithm value of road area per capita (lnpas). The better the level of urban infrastructure, the higher the level of local economic and technological development, so it will affect the degree of resource dependence. We choose per capita road area to represent the level of urban infrastructure according to Yuan et al. [32]. (6) The emission level of pollutants in the city is represented by lnwat [33]. (7) The urban pollutant treatment level, specifically selected as the representative general industrial solid waste comprehensive utilization rate (su), can effectively reflect the local pollution status of an area [34]. (8) The science and technology investment level (sci), which refers to the proportion of science and technology expenditure in government expenditure and the continuous improvement of the regional technology level, will create a good development environment for industrial transformation and technological innovation, thus reducing regional resource dependence [35]. (9) The education input level (edu) refers to the proportion of education expenditure to government expenditure. The education input level can reflect the education status of the local population. The higher the education level, the more support they have for industrial transformation and green development, thus affecting the resource dependence of the region.

3.2.3. Data Description

The data used in this study are all from the <China Urban Statistical Yearbook> from 2006 to 2017. Some of the cities have data missing in some years, which are supplemented by searching the statistical yearbooks of various cities and linear interpolation and finally obtaining the balanced panel data of 113 resource-based cities in China over 12 years. The descriptive statistics of each variable are shown in

Table 1. Descriptive statistics.

Variable	AVG	SD	Min	Max
rd	0.121	0.122	0.010	0.578
lngdp	9.970	0.660	4.425	12.525
lngdpp	99.835	13.192	19.581	156.876
lngov	14.076	0.776	11.325	15.929
lnden	5.402	0.941	2.327	6.941
lnqys	6.089	0.912	3.611	8.331
lnpas	10.733	9.119	0.020	108.370
lnwat	8.096	0.982	4.804	10.288
su	0.732	0.242	0.018	1.350
sci	0.010	0.010	0.000	0.207
edu	0.183	0.043	0.002	0.377

.

4. Results

4.1. Baseline Regression Result

Table 2. Baseline regression result.

	rd
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
DID	−0.0191 ***	−0.0186 ***	−0.0164 ***	−0.0165 ***	−0.0167 ***	−0.0165 ***	−0.0169 ***	−0.0167 ***
	(0.0044)	(0.0044)	(0.0044)	(0.0044)	(0.0044)	(0.0044)	(0.0044)	(0.0044)
lngdp		0.0574 **	0.0646 **	0.0896 ***	0.093 ***	0.094 ***	0.0952 ***	0.0956 ***
		(0.0255)	(0.0254)	(0.0264)	(0.0264)	(0.0264)	(0.0233)	(0.0267)
lngdpp		−0.004 **	−0.0046 ***	−0.0063 ***	−0.0065 ***	−0.0065 ***	−0.0067 ***	−0.0067 ***
		(0.0017)	(0.0017)	(0.0017)	(0.0017)	(0.0017)	(0.0018)	(0.0018)
lnden			0.0926 ***	0.101 ***	0.101 ***	0.1 ***	0.102 ***	0.101 ***
			(0.0226)	(0.0226)	(0.0226)	(0.0226)	(0.0228)	(0.0228)
lnqys				−0.0137 ***	−0.0134 ***	−0.0137 ***	−0.0133 ***	−0.0139 ***
				(0.0041)	(0.004)	(0.004)	(0.0044)	(0.0044)
lnpsa					0.0004 *	0.0004 *	0.0004 *	0.0004
					(0.0002)	(0.0002)	(0.0002)	(0.0002)
lnwat						0.0039 *		0.0039 *
						(0.002)		(0.002)
lngov							−0.0027	−0.0017
							(0.0078)	(0.0078)
sci							0.0225	0.0379
							(0.11)	(0.11)
edu							−0.0333	−0.0274
							(0.0395)	(0.0396)
su						−0.0062		−0.0060
						(0.0062)		(0.0063)
cons	0.127 ***	−0.057	−0.5677 ***	−0.615 ***	−0.6316 ***	−0.6629 ***	−0.6053 ***	−0.646 ***
	(0.0028)	(0.1014)	(0.1603)	(0.1602)	(0.1603)	(0.1608)	(0.1763)	(0.1771)
Year fixed effect	Y	Y	Y	Y	Y	Y	Y	Y
Individual fixed effect	Y	Y	Y	Y	Y	Y	Y	Y
Number of samples	1356	1356	1356	1356	1356	1356	1356	1356
R2	0.191	0.194	0.205	0.213	0.215	0.218	0.215	0.281

Notes: 1. The values in brackets represent robust standard errors of the estimated coefficients, with *, **, and *** representing significance levels of 10%, 5%, and 1%, respectively. 2. Y: Yes.

shows the baseline regression results of Formula (1), where column (1) is the regression results without control variables, and columns (2)–(8) are the regression results with the gradually introducing control variables. The results show that in the process of gradually introducing control variables, the DID coefficients are all significantly negative at the 1% level. When there are no control variables, the coefficient value is −0.0148, and after introducing all control variables, the coefficient value is −0.0133, which indicates that the implementation of the transfer payment policy has significantly reduced the resource dependence of resource-depleted cities. Each additional unit of transfer payment can reduce the proportion of workers in the extractive industry to 1.67% of the total population at the end of the year; that is, the implementation of a transfer payment policy can reduce the number of workers in the extractive industry, thus reducing the degree of local resource dependence. Therefore, Hypothesis 1 is confirmed.

From the perspective of control variables, in columns (2)–(8), the per capita GDP terms are positively correlated with resource dependence, while the quadratic terms of the per capita GDP are negatively correlated. This shows that with the improvement of the level of economic development, the degree of dependence on resources has a trend of first rising and then falling, that is, an inverted “U”-shaped relationship. The absolute value of the quadratic term coefficient is small, indicating that it is a very long process to rely on the improvement of economic development to reduce local resource dependence. The coefficient of population density terms in columns (3)–(8) are positive, which means the improvement of the urbanization level will increase local dependence on resources. This may be due to the coexistence of the population agglomeration dividend brought about by urbanization and the social survival pressure brought about by population increases, but the pressure is significantly greater than the demographic dividend, which makes the development of resource-based cities more dependent on the exploitation of local resources. The coefficients of the number of industrial enterprises above the designated size in columns (4)–(8) are significantly negative, indicating that the increase in the number of urban industrial enterprises is conducive to reducing the dependence on resources. This is because the industrial enterprises above the designated size are not only mining enterprises that rely on natural resources but also assembly-type and modulation-type enterprises that do not rely on natural resources for development. The increase in the number of such enterprises can share the pressure of the employment rate of some state-owned mining enterprises, thereby reducing the dependence on resources. This result preliminarily shows that the transformation of industrial structures has a positive effect on resource dependence. The coefficients of per capita urban road area are significantly positive in columns (5)–(7), but the value is small, indicating that the level of urban infrastructure construction has little impact on resource dependence. The effect of the remaining control variables is not obvious, so they will not be explained.

4.2. Parallel Trend Test

To test the hypothesis of parallel trends and analyze the dynamic effects of policies, we referred to the practice of Wu et al. [31] and conducted the test based on event analysis. In order to reflect the implementation of the three transfer payment policies at different times better, we draw a time trend graph for each policy implementation separately.

Firstly, China identified the first batch of resource-depleted cities in 2008 and proposed to continue to provide transfer payments to resource-depleted cities except for Panjin city in 2011, so here we consider the first batch of resource-depleted cities as having received two policy shocks. We exclude data from the second and third batches of resource-depleted cities as well as Panjin city after 2011, and plot Figure 1 only uses the first batch of resource-depleted cities as the experimental group and the remaining resource-based cities as the control group. The diagnostic statistics of the first batch of resource-depleted cities (experimental group) and other cities (control group) are shown in

Table 3. Diagnostic statistics of the first batch of resource-depleted cities.

Year	Experimental Group	Control Group
2006	0.1462222	0.1256154
2007	0.144	0.1232596
2008	0.1571111	0.1298942
2009	0.159	0.1299904
2010	0.1592222	0.1301635
2011	0.1571111	0.1333365
2012	0.1565555	0.1312308
2013	0.1255556	0.1224038
2014	0.1245556	0.1140192
2015	0.1171111	0.1044135
2016	0.0996667	0.1002885
2017	0.0956667	0.0952404

.

Table 3 and Figure 1 present that the policy in 2008 began to show effects in 2010, and the resource dependence of the experimental group began to decline. The 2011 policy showed a completely different trend from the experimental group during the implementation period, with a slight increase at first and then a significant decrease. The effect of the policy in 2008 was weak and accompanied by a delay. On the one hand, the first batch of resource-depleted cities faced high learning costs, and there was no previous successful transformation experience to learn from. The road to transformation is still in the exploratory stage, so the policy effect is not obvious. On the other hand, as the first batch of resource-depleted cities faced the most serious urban problems, with serious industrial consolidation and a high proportion of secondary industries, resource-dependent enterprises were unable to complete the transformation and upgrading of their own industries in a short period, thus creating a policy of delay.

Secondly, China identified the second batch of resource-depleted cities in 2009. We exclude data from the first and third batches of resource-depleted cities, and plot Figure 2 only uses the second batch of resource-depleted cities as the experimental group and the remaining resource-based cities as the control group. The diagnostic statistics of the first batch of resource-depleted cities (experimental group) and other cities (control group) are shown in

Table 4. Diagnostic statistics of the second batch of resource-depleted cities.

Year	Experimental Group	Control Group
2006	0.1996667	0.1131348
2007	0.1950667	0.1111573
2008	0.1968667	0.1186067
2009	0.1970667	0.1186854
2010	0.2001333	0.1183708
2011	0.2094667	0.1205056
2012	0.2051333	0.1187753
2013	0.1796667	0.1127528
2014	0.1659333	0.1052697
2015	0.1454	0.0975056
2016	0.1303333	0.0952247
2017	0.1205333	0.0909775

.

From Table 4 and Figure 2, it can be seen that after the implementation of the policy, the trend of the experimental group is no longer the same as that of the control group, with a slight increase at first and then a downward trend since 2011. Firstly, the slight increase may be due to the fact that the enterprises are dealing with the closing of the extractive industry, and some of the holes that have been developed in the early stage need to continue to be mined in a certain amount to reduce the development loss. Secondly, the policy effect of the second batch of resource-depleted cities is also temporarily delayed, but the delay was shorter compared to the first batch of resource-depleted cities. This is because there are precedents of the first batch of resource-depleted cities that can be used for reference, so the second batch of resource-depleted cities has a relative advantage in the transition.

In addition, China identified the third batch of resource-depleted cities in 2012. We plot Figure 3 to show the trend of the third batch of resource-depleted cities and other resource-based cities. The diagnostic statistics of the first batch of resource-depleted cities (experimental group) and other cities (control group) are shown in

Table 5. Diagnostic statistics of the third batch of resource-depleted cities.

Year	Experimental Group	Control Group
2006	0.1431429	0.1131348
2007	0.1374286	0.1111573
2008	0.1348571	0.1186067
2009	0.1418571	0.1186854
2010	0.1497143	0.1183708
2011	0.1795714	0.1205056
2012	0.1694286	0.1187753
2013	0.1542857	0.1127528
2014	0.129	0.1052697
2015	0.0884286	0.0975056
2016	0.0791429	0.0952247
2017	0.0711429	0.0909775

. It can be seen from Table 5 and Figure 3 that the resource dependence of the experimental group appears to rise since the first batch of resource-depleted cities was identified in 2008, while the resource dependence started to decrease after the declaration process ended in 2011. The rise is due to the intentional upgrading of resource extraction by the cities concerned in the hope of passing the audit of the third resource-depleted cities in order to obtain the transfer payments issued by the central government. After the declaration ended and the policy started to be implemented, the cities started to focus on industrial transformation efforts. The decline in resource dependence here is a joint effect of the return to normal levels of extraction and the policy itself. Based on the above analysis, we believe that the data used in the article satisfy the parallel trend test.

Next, to further examine parallel trends, we conduct a dynamic effect analysis based on the event study method. Due to the large difference in the time points of the three divisions of resource-depleted cities, the number of years before and after the implementation of the policy is different. Therefore, we analyze the dynamic effects of the three policies, respectively. We construct the following model using the year after policy implementation as a comparison benchmark.

$$r{d}_{it}={\alpha }_0+{\displaystyle \sum }_s{\alpha }_{pr{e}_s}{D}_{pr{e}_s}+{\alpha }_{policy}{D}_{policy}+{\displaystyle \sum }_{s\ne 1}{\alpha }_{pos{t}_s}{D}_{pos{t}_s}+{\eta }_i+{\gamma }_t+{\epsilon }_{it}$$

(2)

In Formula (2), $D_{pr{e}_s}, D_{policy}, D_{pos{t}_s},$ respectively, represent the multiplication term of the year dummy variable and the corresponding policy dummy variable before, when, and after the policy implementation, and ${\alpha }_{pr{e}_s}, {\alpha }_{policy}, {\alpha }_{pos{t}_s}$ is the corresponding coefficient. If it is positive, it means that the influence of the policy is positive; otherwise, the influence is negative. The regression results are shown in

Table 6. The result of parallel trend test.

	rd
	2008		2009		2012
	(1)	(2)	(3)	(4)	(5)	(6)
pre4					−0.0164	−0.0156
					−0.013	−0.0128
pre3			0.0048	0.0012	−0.0094	−0.0101
			−0.0115	−0.0113	−0.013	−0.0128
pre2	0.0024	−0.0012	0.0021	0.0016	−0.0013	−0.0039
	−0.0097	−0.0095	−0.0115	−0.0113	−0.0131	−0.0127
pre1	0.0006	−0.0011	−0.0035	−0.0027	0.0265	0.0249
	−0.0096	−0.0094	−0.0114	−0.0112	−0.013	−0.0128
policy	−0.0007	−0.0013	−0.0034	−0.0026	0.018	0.0153
	−0.0096	−0.0095	−0.0114	−0.0114	−0.013	−0.0131
post2	0.0023	0.0021	0.0072	0.008	−0.0089 *	−0.0101 *
	−0.0095	−0.0094	−0.0115	−0.0113	−0.013	−0.0128
post3	0.0052	0.0052	0.0045	0.0045	−0.0416 ***	−0.0444 ***
	−0.0096	−0.0094	−0.0116	−0.0115	−0.0131	−0.0129
post4	0.004	0.0032	−0.0148 *	−0.0128 **	−0.0487 ***	−0.0513 ***
	−0.0096	−0.0095	−0.0115	−0.0113	−0.013	−0.0129
post5	−0.0175 **	−0.0156 **	−0.0211 ***	−0.0185 ***	−0.0524 ***	−0.0486 ***
	−0.0095	−0.0093	−0.0114	−0.0112	−0.0129	−0.013
post6	−0.0189 ***	−0.0178 ***	−0.0339 ***	−0.0353 ***
	−0.0094	−0.0093	−0.0114	−0.0112
post7	−0.0268 ***	−0.0288 ***	−0.0467 ***	−0.0482 ***
	−0.0094	−0.0095	−0.0114	−0.0114
post8	−0.0405 ***	−0.0425 ***	−0.0522 ***	−0.0501 ***
	−0.0094	−0.0095	−0.0114	−0.0113
post9	−0.0438 ***	−0.0443 ***
	−0.0094	−0.0096
Control variables	N	Y	N	Y	N	Y
Year fixed effect	Y	Y	Y	Y	Y	Y
Individual fixed effect	Y	Y	Y	Y	Y	Y
Number of samples	1176	1176	1164	1164	1152	1152
R2	0.227	0.255	0.217	0.254	0.187	0.216

Notes: 1. The values in brackets represent robust standard errors of the estimated coefficients, with *, **, and *** representing significance levels of 10%, 5%, and 1%, respectively. 2. Y: Yes, N: No.

, and Figure 4, Figure 5 and Figure 6 are drawn according to the regression results.

Table 6 shows the value of the regression coefficient ${\alpha }_{pr{e}_s}, {\alpha }_{policy}, {\alpha }_{pos{t}_s}$ in different batches. Figure 4, Figure 5 and Figure 6 show the dynamic effect coefficient of three batches of resource-depleted cities. Since our data interval is from 2006 to 2017, we can see the effect of two years before and nine years after the establishment of the first batch of resource-depleted cities in 2008 (here we consider the 2011 policy as a supplement to the 2008 policy, and it will not be discussed separately).

The second and third batches of resource-depleted cities are excluded from the data, and the dynamic effect analysis is carried out on the first batch of resource-depleted cities. The results are shown in columns (1) and (2) in Table 6 and the dynamic effect coefficient is shown in Figure 4. We can see that the coefficients before the policy implementation are not significant, and the coefficients are significantly negative from the 5th year after the policy implementation, indicating that the data pass the parallel trend test. It should be noted here that the results show that the policy effect in 2008 was delayed for a long time, indicating that the effect of the initial implementation of transfer payments is not ideal. This corroborates the statement of why China introduced the continuation of transfer payments to the first batch of resource-depleted cities in 2011. At the same time, it also shows that our data results are highly consistent with reality.

From columns (3) and (4) in Table 6 and Figure 5, we can see the effect three years before and eight years after the establishment of the second batch of resource-depleted cities in 2009. The coefficients before the policy implementation are not significant, and the coefficients are significantly negative from the 4th year after the policy implementation, so the data also pass the parallel trend test. Compared with the 2008 policy, the 2009 policy delay is shorter, indicating that the second batch of resource-depleted cities is more sensitive and active in responding to the transfer payment policy.

From columns (5) and (6) in Table 6 and Figure 6, we can see the effect of four years before and five years after the establishment of the second batch of resource-depleted cities in 2012. After the implementation of this policy, the coefficient has been significantly negative, and there is no delay effect. This shows that the third batch of resource-depleted cities already have certain expectations and plans for transformation, so they respond very quickly to policies. Combining the above figures and table, we further illustrate that the data used in the article pass the parallel trend test.

4.3. Robustness Test

4.3.1. Propensity Score Matching and Difference-in-Differences Analysis

Referring to Shi et al., this study uses a propensity score matching and difference-in-differences model to test the robustness of the baseline regression results and uses radius matching, proximity matching, and kernel matching methods to verify [63]. Considering that the earliest implementation year of the policy was 2008, the policy may affect the changes in the relevant economic variables in the pilot areas [31], so this study carries out propensity score matching on the samples only in 2006 and 2007. We use propensity score matching to find the control group with the characteristics closest to the experimental group and retain the sample points within the common value range in the aforementioned two years. The matched experimental group and control group are used for regression. The specific model is as follows:

$$r{d}_{it}^{PSM}={\beta }_0+{\beta }_{1}DI{D}_{it}+{\beta }_{2}contro{l}_{it}+{\eta }_i+{\gamma }_t+{\epsilon }_{it}$$

(3)

The regression results are shown in

Table 7. Robustness test of propensity score matching.

	rd
	Radius Matching	Proximity Matching	Kernel Matching
DID	−0.0160 ***	−0.0146 ***	−0.0154 ***
	(0.0049)	(0.0046)	(0.0048)
Control variables	Y	Y	Y
Year fixed effect	Y	Y	Y
Individual fixed effect	Y	Y	Y
Number of samples	876	924	900
R2	0.251	0.26	0.253

Notes: 1. The values in brackets represent robust standard errors of the estimated coefficients, with *** representing significance levels of 1%. 2. Y: Yes.

. No matter what matching method is adopted, the regression coefficient of the DID terms is negative at the significance level of 1%, and the value is relatively close to that obtained in the baseline regression. Therefore, the baseline regression results in this study are robust.

4.3.2. Placebo Test

To further verify the robustness of the baseline regression results, a placebo test is performed. First, we carry out random sampling on all resource-based cities and policy time within the sample, and 24 prefecture-level resource-based cities and their corresponding random policy implementation time points are selected each time; that is, 24 cities correspond to 24 policy implementation time points. Then, take the 24 cities as the virtual experimental group and the remaining cities as the virtual control group for regression.

Repeat the above process 500 times to obtain the DID regression coefficients of interaction between 500 virtual processing groups and virtual policy times (see Figure 7). In Figure 7, the regression coefficients of 500 times are mainly concentrated around 0, and the p values of most estimated values are greater than 0.1; that is, they are not significant at the significance level of 10%. Meanwhile, the vertical dotted line of the reference regression results is shown in the figure, which is located at the low tail of the distribution of regression coefficients, indicating that the reference regression results are not obtained by chance. Figure 7 illustrates that the results pass the placebo test; that is, the baseline regression results are robust.

4.3.3. Change the Explained Variables

To eliminate the statistical bias, robustness tests are carried out by using alternative explained variables. In the baseline regression Formula (1), the explained variable $Y_{it}$ is calculated as the ratio of the number of employees in the mining industry to the total urban population. We use the ratio of the number of employees in the mining industry to the number of employees at the end of the year (Alter1) and the ratio of the number of employees in the mining industry to the number of employees in the secondary industry (Alter2) to eliminate the error caused by variable selection. After substituting the dependent variable, the regression is performed according to the following model:

$$alte{r}_{it}={\beta }_0+{\beta }_{1}DI{D}_{it}+{\beta }_{2}contro{l}_{it}+{\eta }_i+{\gamma }_t+{\epsilon }_{it}$$

(4)

The results are shown in column (1)–(4) in

Table 8. Change the explained variables.

	Alter1		Alter2		rd
	(1)	(2)	(3)	(4)	(5)	(6)
DID	−0.0191 ***	−0.0162 ***	−0.0351 ***	−0.0349 ***	−0.0148 ***	−0.0133 ***
	−0.0041	−0.0041	−0.0087	−0.0088	−0.005	−0.0051
Control variables	N	Y	N	Y	N	Y
Year fixed effect	Y	Y	Y	Y	Y	Y
Individual fixed effect	Y	Y	Y	Y	Y	Y
Province individual time effect	N	N	N	N	Y	Y
Number of samples	1356	1356	1356	1356	1356	1356
R2	0.194	0.226	0.221	0.228	0.361	0.379

Notes:1. The values in brackets represent robust standard errors of the estimated coefficients, with *** representing significance levels of 1%. 2. Y: Yes, N: No.

. It can be found that whether the control variables are added or not, the regression coefficient of the DID term is negative at a 1% significance level, indicating that the transfer payment policy can reduce the resource dependence of resource-depleted cities. The baseline regression results are robust.

4.3.4. Consider Province-Year Fixed Effect

In the baseline regression Formula (1), we only consider city-individual fixed effects and time fixed effects, but there are still many unobservable factors that will affect the regression results. Referring to Li and Shi [20,63], we add a province-year fixed effect to control the effect of unobservable factors that vary over time in each city. The specific model is as follows:

$$r{d}_{it}={\beta }_0+{\beta }_{1}DI{D}_{it}+{\beta }_{2}contro{l}_{it}+{\eta }_i+{\gamma }_t+{\delta }_t^j+{\epsilon }_{it}$$

(5)

wherein ${\delta }_t^j={\mathrm{Province}}_j\times Yea{r}_t$, ${\mathrm{Province}}_j$ represents the jth province and $Yea{r}_t$ represents the tth year. The regression result is shown in column (5)–(6) in Table 8. It can be seen from the results that the coefficient of the DID term is still significantly negative after the provincial-year effect is fixed, indicating that the transfer payment policy can effectively reduce the resource dependence of resource-depleted cities, and it also shows that the baseline regression results are robust.

4.3.5. Exclude Other Policy Interference

To further test the robustness of baseline regression, the effects of other relevant policies during the same period are considered. Since other environmental policies in the same period may also affect the resource dependence of resource-depleted cities, the results obtained by the baseline regression may be formed by multiple policies. The low-carbon city pilot policy was first implemented in 2010, and the carbon market pilot policy was first implemented in 2012. Both policies can reduce carbon emissions in the corresponding pilot areas. The exploitation and processing of natural resources will generate a large amount of carbon dioxide, so the low-carbon city pilot policy and carbon market pilot policy will inhibit the exploitation of natural resources to a certain extent, which can interfere with the baseline regression results. Although the <Air Pollution Prevention and Control Action Plan> enacted in 2013 and other environmental policies will have an impact on the exploitation of local natural resources, these policies affect all cities in China and cannot be excluded from our research. The selection of resource-based cities in the research scope of our research has eliminated the endogeneity problem to the greatest extent. After excluding low-carbon city pilot policy and carbon market pilot policy areas, the regression is performed according to Formula (1), and the results are shown in

Table 9. Exclude Other Policy Interference.

	rd
	Exclude Low-Carbon Pilot Cities	Exclude Low-Carbon Pilot Cities	Exclude Carbon Market Pilots	Exclude Carbon Market Pilots	Exclude Low-Carbon Pilot Cities and Carbon Market Pilots	Exclude Low-Carbon Pilot Cities and Carbon Market Pilots
DID	−0.0199 ***	−0.0158 ***	−0.0195 ***	−0.0172 ***	−0.0204 ***	−0.0164 ***
	(0.0047)	(0.0047)	(0.0048)	(0.0048)	(0.0051)	(0.0051)
lngdp		0.19 ***		0.0974 ***		0.193 ***
		(0.061)		(0.0277)		(0.0635)
lngdpp		−0.011 ***		−0.0068 ***		−0.0111 ***
		(0.0029)		(0.0019)		(0.0031)
lnden		0.103 ***		0.0991 ***		0.101 ***
		(0.0243)		(0.0236)		(0.0252)
lnqys		−0.016 ***		−0.014 ***		−0.0165 ***
		(0.0048)		(0.0046)		(0.0050)
lnpsa		0.0004 *		0.0004 *		0.0005 *
		(0.0002)		(0.0002)		(0.0002)
lnwat		0.0033		0.0041		0.0035
		(0.0022)		(0.0021)		(0.0023)
lngov		−0.003		−0.0049		−0.0057
		(0.0089)		(0.0082)		(0.0095)
sci		0.0534		0.0811		0.106
		(0.115)		(0.149)		(0.156)
edu		−0.0308		−0.0386		−0.0427
		(0.0436)		(0.0418)		(0.0458)
su		−0.0074		−0.0038		−0.0053
		(0.0067)		(0.0066)		(0.007)
cons	0.130 ***	−1.125 ***	0.132 ***	−0.593 ***	0.134 ***	−1.088 ***
	(0.0031)	(0.319)	(0.003)	(0.183)	(0.0032)	(0.327)
Year fixed effect	Y	Y	Y	Y	Y	Y
Individual fixed effect	Y	Y	Y	Y	Y	Y
Number of samples	1236	1236	1272	1272	1164	1164
R2	0.192	0.224	0.187	0.215	0.187	0.22

Notes: 1. The values in brackets represent robust standard errors of the estimated coefficients, with * and *** representing significance levels of 10% and 1%, respectively. 2. Y: Yes.

. It can be seen from the results that the implementation of a transfer payment policy can still significantly reduce the resource dependence of resource-depleted cities after excluding the role of low-carbon city pilot policies and carbon market policies, so the baseline regression results are robust.

4.4. Heterogeneity Analysis

To verify Hypothesis 2 and examine regional heterogeneity, the city samples are divided into the eastern region, the central region, and the western region. The eastern region contains 28 resource-based cities including 5 resource-depleted cities, the central region contains 55 resource-based cities including 15 resource-depleted cities, and the western region contains 30 resource-based cities including 4 resource-based cities. The regression shown in Formula (1) is carried out under the sample ranges of eastern, central, and western regions, respectively. The results are shown in

Table 10. Regional heterogeneity.

	Geographical Location			City Size			Resource Type
	Eastern	Central	Western	Small and Medium	Large	Mega	Coal	Forestry	Metal	Petroleum
DID	0.0016	−0.0181 ***	−0.0252 **	−0.0625 ***	−0.0148 ***	0.0146	−0.0218 ***	−0.0011	−0.0045	−0.0326 *
	−0.0065	−0.0063	−0.0124	−0.0199	−0.0054	−0.0111	−0.006	−0.0179	−0.0059	−0.0153
Control variables	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Year fixed effect	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Individual fixed effect	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Number of samples	336	660	360	84	960	312	984	72	192	108
R2	0.383	0.275	0.173	0.646	0.195	0.363	0.273	0.458	0.352	0.51

Notes: 1. The values in brackets represent robust standard errors of the estimated coefficients, with *, **, and *** representing significance levels of 10%, 5%, and 1%, respectively. 2. Y: Yes.

.

It can be seen from the geographical location columns that the regression coefficients of the DID term in the central and western regions are negative at a 5% significance level, and the numerical values are less than the baseline regression results. This indicates that the implementation of the transfer payment policy in central and western regions can effectively help local resource-depleted cities to reduce resource dependence, and this effect is more prominent in the western region. The regression coefficient in the eastern region is not significant and is close to 0, which to a certain extent indicates that the transfer payment policy is not ideal for resource-depleted cities in the eastern region. The differences in original resource endowment and economic and social development in eastern, central, and western regions result in different responses to the transfer payment policy. The eastern regions are relatively developed, and, overall, industrial structures are advanced and complete. Moreover, the eastern region has more social resources such as funds, talents, and infrastructure, so the marginal effect of the transfer payment policy on reducing local resource dependence is small, or even insufficient. On the contrary, the central and western regions are relatively backward in terms of economic development, social development, and urban infrastructure construction. The transfer payment policy can play a better role in the structural upgrading of enterprises, green technological innovation, and the reduction of resource exploitation. From the regression results, it can be seen that the western region has the largest policy role space. So far, Hypothesis 2 has been verified.

To explore more heterogeneity, we considered the effect of heterogeneity on city size and resource type on the degree of resource dependence. The difference in city size means that each city has a different economic agglomeration effect and different efficiency in resource allocation and utilization, which ultimately leads to a larger difference in the degree of response to transfer policies. The 113 resource-based cities in the article are divided in conjunction with the <Notice on Adjusting the Criteria for the Division of Urban Scale> issued by the State Council in November 2014. Since the State Council divides cities into five categories, among which the samples of small cities and super-large cities are scarce, we refer to Li and Li to integrate cities into three categories, small and medium-sized cities, large cities, megacities (super and super-large cities) [64]. The regression results are shown in the city size column in Table 10. We find that the DID terms of small and medium-sized cities and large cities are significantly negative at the 1% level, while the regression coefficients of megacities are not significant. Since only one city was identified as a resource-depleted city in the megacity category, the results obtained are not universal, and more attention should be paid to the results of small and medium-sized cities and large cities. The results show that small and medium-sized cities have a stronger effect of reducing resource dependence than large cities, which is counter-intuitive. The common perception is that large cities have stronger economic agglomeration effects than small cities and should respond more positively to policies so that there should be a greater degree of reduction in resource dependence. By comparing the number of employees per unit of industrial enterprises in the two types of cities, we believe that the reason for this result is that in the process of industrial transformation, large cities will face more corporate personnel flows, such as layoffs, job transfer, and re-employment after training, etc. On the one hand, frequent staff turnover is a big cost for enterprises. On the other hand, and more importantly, the unemployment rate is a key concern for the local government, and many local extractive enterprises are state-owned enterprises facing the situation of being unable to lay off staff and the difficulty of transformation. Small and medium-sized cities face relatively low silent costs in the process of industrial structure transformation, so their response to transfer payment policies is more active and effective than large cities.

The main resources of different resource-depleted cities are different, which will also make them respond differently to the effect of transfer payment policies. Referring to Li and Wang [22], we divide resource-based cities into four categories according to their resources: coal, forestry, metal, and petroleum. Forestry cities account for a minority of resource-based cities, with resources mainly in forests and trees, and the regions are distributed in Jilin, Heilongjiang, and Yunnan. Metal city resources include non-ferrous and ferrous metals, mostly distributed in Liaoning, Henan, Jiangxi, and Gansu provinces. The regression results are shown in the resource type column in Table 10. The results show that the DID term is significantly negative for both coal and oil types, indicating that the transfer payment policy has a positive effect on resource-depleted cities with coal and oil as the main extracted resources. Meanwhile, the regression coefficients of the DID coefficients for the two resources of forestry and metals fail to pass the significance test, although they are negative, indicating that resource-depleted cities dominated by these two resources do not respond significantly to the transfer payment policy. The reason for this result is that coal resources are mostly used for energy conversion after mining, such as electricity and heating. The current research and exploration of new energy generation is becoming more and more mature, which to some extent relieves the demand for coal resources in various regions. This provides a positive role in the transformation of resource-depleted cities with coal mining as the mainstay industry, so the effect of the transfer payment policy is most significant. In contrast, wood is mostly used for building houses and making furniture after mining, while metals are mostly used for manufacturing large machines and precision instruments. Due to the irreplaceable nature of wood and metal resources, the demand for both resources remains high, and resource-depleted cities face the dual pressure of meeting market demands and addressing resource depletion. Therefore, resource-depleted cities with wood and metal as their core industries face more restrictions and challenges in the process of industrial transformation than coal-based resource-depleted cities, which makes the transfer payment policy not so effective.

The development of China’s resource-depleted cities faces some dilemmas. If the economy continues to rely on resources, it will be completely depleted and the city will also face more serious air pollution problems. If we want to get rid of the dependence on resources, we must change the mode of production, which requires a lot of capital investment, talent introduction, and technological innovation, and the change in the mode of production will make workers face the risk of unemployment. According to the research results of this study, the empirical study of the impact of transfer payments on the resource dependence degree is conducted by using the difference-in-differences model. It proves that the transfer payment policy can help reduce the resource dependence of resource-depleted cities, which plays an important guiding role in the financial policy governance of resource-depleted cities in China. At the same time, according to the results of Hypothesis 2, the effects of transfer payment policies have regional heterogeneity, and the eastern, central, and western regions will have different responses to transfer payment policies, which also puts forward different challenges and requirements for the financial policies of resource-depleted cities in different regions of China. The eastern, central, and western regions need to take appropriate financial measures in light of their actual conditions to reduce the dependence of their cities on resources, which is of great economic significance for maintaining energy security, safeguarding local people’s livelihood, and ensuring high-quality and sustainable economic development.

4.5. Mechanism Analysis

In order to investigate the mechanism of transfer payment policy on resource dependence of resource-depleted cities, this study introduces the rationalization of industrial structures (RIS) and the optimization of industrial structures (OIS) as mechanism variables from the path of regional industrial structure transformation, the government expenditure on science and technology education (INF) as a mechanism variable from the path of infrastructure construction, and technological progress (TEC) as a mechanism variable from the path of improving green innovation, thereby improving production efficiency. The rationalization of industrial structures is used to measure the degree of aggregation between different industries. The optimization of industrial structures is used to measure the upgrading of industrial structures [62]. The expenditure on education of science and technology is used to measure the level of infrastructure construction in the region, and technological progress is used to measure the improvement of resource utilization efficiency. Drawing on the research method of Wu et al., we first examine the influence of policy and time multiplication terms (DID) on the mechanism variable and then examine the influence of the mechanism variable on the explained variable [31]. The specific model is as follows:

$$ME{C}_{it}={\theta }_0+{\theta }_{1}DI{D}_{it}+{\theta }_{2}contro{l}_{it}+{\eta }_i+{\gamma }_t+{\epsilon }_{it}$$

(6)

$$r{d}_{it}={\beta }_0^{\prime }+{\beta }_1^{\prime }DI{D}_{it}+{\beta }_2^{\prime }DI{D}_{it}\times ME{C}_{it}+{\beta }_3^{\prime }ME{C}_{it}+{\beta }_4^{\prime }contro{l}_{it}+{\eta }_i+{\gamma }_t+{\epsilon }_{it}$$

(7)

wherein $ME{C}_{it}$ represents the mechanism variable of city $i$ in the $t$th year. When $ME{C}_{it}$ is $RI{S}_{it},$ it represents the rationalization of the industrial structure of city $i$ in the $t$th year. When it is $OI{S}_{it,}$ it represents the optimization of the industrial structure of city $i$ in the $t$th year. When it is $IN{F}_{it}$, it represents the infrastructure construction level of city $i$ in the $t$th year. When it is $TE{C}_{it}$, it represents the technological progress of city $i$ in the $t$th year. The calculation method of the index is as follows:

$$RIS={\displaystyle \sum }_{i=1}^3(\frac{Y_i}{Y})\ln (\frac{Y_i}{L_i}/\frac{Y}{L})$$

(8)

$$OIS=\frac{Y_3}{Y_2}$$

(9)

The INF variable is represented by the logarithmic value of government expenditure on science, technology, and education, and the TEC variable is represented by the sewage treatment rate. As shown in Formulas (8) and (9), $Y_i$ represents the output value of the $i$th industry, $L_i$ represents the number of employees in the $i$th industry, and $Y$ and $L$ represent the region’s total output value and the total number of employees, respectively. The calculation of the rationalization of the industrial structures adopts the Theil index measurement method. According to economic theory, when the economy is in the final equilibrium state, the production efficiency between the industries should be the same (namely, $Y_i/L_i=Y/L$, and the closer RIS is to 0, the more stable the aggregation between the industries is). In order to facilitate the analysis of the directionality of the effect, the absolute value of RIS is used in the corresponding regression. The optimization of industrial structures is measured by the ratio of the GDP of the tertiary industry to the GDP of the secondary industry. The higher the OIS value, the more important the tertiary industry is in the local economic development.

When $ME{C}_{it}$ is $RI{S}_{it}$, the regression coefficient ${\theta }_1$ in Formula (6) reflects the influence of transfer payment policy on the rationalization of industrial structure. If it is negative, it indicates that the transfer payment policy helps to promote coordination between various industries and makes the local industrial structures more stable and reasonable. In Formula (7), $DI{D}_{it}\ast RI{S}_{it}$ is an interaction term and the coefficient ${\beta }_1^{\prime }$ represents the effect of transfer payment policy on resource dependence through rationalization of industrial structure. If it is negative, it means that the transfer payment policy can help enhance resource dependence reduction in resource-depleted cities by coordinating the resource utilization efficiency of various industries. When $ME{C}_{it}$ is $OI{S}_{it}$, the regression coefficient ${\theta }_1$ in Formula (6) reflects the influence of transfer payment policy on the optimization of industrial structure. If it is positive, it indicates that the transfer payment policy helps to promote the transformation of the local industrial structures from the secondary industry to the tertiary industry; otherwise, it shows a negative impact. In Formula (7), $DI{D}_{it}\ast OI{S}_{it}$ is an interaction term and the coefficient ${\beta }_1^{\prime }$ represents the degree to which the transfer payment policy affects resource dependence through the optimization of industrial structures. If it is negative, it means that the transfer payment policy can help resource-depleted cities to reduce resource dependence by means of industrial structure transformation at an advanced level. When $ME{C}_{it}$ is $IN{F}_{it}$, the regression coefficient ${\theta }_1$ in Formula (6) reflects the impact of the transfer payment policy on the infrastructure construction level. If it is positive, it indicates that the policy has a positive impact on local infrastructure construction; otherwise, it has a negative impact. In Formula (7), $DI{D}_{it}\ast IN{F}_{it}$ is an interaction term, and if the coefficient ${\beta }_1^{\prime }$ is negative, it means that the transfer payment policy can reduce the resource dependence of resource-depleted cities by strengthening infrastructure construction. When $ME{C}_{it}$ is $TE{C}_{it}$, the regression coefficient ${\theta }_1$ Formula (6) reflects the impact of the transfer payment policy on technological progress; otherwise, it has a negative impact. In Formula (7), $DI{D}_{it}\ast TE{C}_{it}$ is an interaction term, and if the coefficient ${\beta }_1^{\prime }$ is negative, it means that the transfer payment policy can reduce the resource dependence of resource-depleted cities through technological progress. Regression is performed according to Formulas (6) and (7), and the results are reported in

Table 11. Mechanism analysis.

	RIS	OIS	INF	TEC	rd
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
DID	2.4422 *	0.0529 ***	−0.0246	−0.0278	−0.0153 **	−0.001	0.0534 ***	0.2188 ***
	(1.3554)	(0.0259)	(0.0091)	(0.0206)	(0.0063)	(0.0067)	(0.013)	(0.0625)
DID $\times$ RIS					−0.0002 *
					(0.0002)
DID $\times$ OIS						−0.0223 ***
						(0.0073)
DID $\times$ INF							−0.0188 ***
							(0.0051)
DID $\times$ TEC								−0.089 ***
								(0.0155)
Control variables	Y	Y	Y	Y	Y	Y	Y	Y
Year fixed effect	Y	Y	Y	Y	Y	Y	Y	Y
Individual fixed effect	Y	Y	Y	Y	Y	Y	Y	Y
Number of samples	1356	1356	1356	1356	1356	1356	1356	1356
R2	0.048	0.488	0.991	0.522	0.228	0.225	0.229	0.239

Notes: 1. the values in brackets represent robust standard errors of the estimated coefficients, with *, **, and *** representing significance levels of 10%, 5%, and 1%, respectively. 2. Y: Yes.

.

Columns (1)–(4) in Table 11 show the results of the impact of the transfer payment policy on mechanism variables, and columns (5)–(8) show the estimated results of interactive items between the transfer payment policy and the mechanism variables’ effects on resource dependence. It can be seen that the regression coefficient in column (1) is positive, which means that the transfer payment policy has a negative effect on the rationalization of industrial structure; that is, it reduces the coordination degree between different industrial structures in resource-depleted cities. Regression coefficients in column (2) are positive at the 1% significance level, showing that the transfer payment policy plays a significant role in promoting the industrial structure optimization index. This shows that the optimization of industrial structures is conducive to the evolution of the secondary industry to the tertiary industry in resource-depleted cities, and it also shows that the evolution of the industrial structures to the advanced level sacrifices the coordination between different industries. The regression coefficients in columns (3) and (4) are not significant, indicating that the effect of the transfer payment policy on local infrastructure construction and technological progress is not obvious. Furthermore, the regression coefficients of the interaction terms in columns (5)–(8) are all significantly negative, which shows that transfer payment policy can reduce the resource dependence of resource-depleted cities by promoting the transformation of the regional industrial structures to a more rational and advanced level, strengthening the construction of local infrastructure levels and improving the efficiency of resource utilization. Through numerical analysis, the reduction of resource dependence by means of the rationalization of industrial structures is very small, so it can be concluded that the industrial structure transformation is mainly based on optimization and supplemented by rationalization. Based on the above analysis, it can be concluded that the optimization of industrial structures, infrastructure construction, and technological progress is the primary way that the transfer payment policy can reduce resource dependence, and the rationalization of industrial structures is the secondary way. So far, Hypothesis 3 has been verified.

5. Discussion

On the one hand, the transfer payment policy brings more opportunities to enterprises in resource-depleted cities. Whether it provides tax subsidies or technical assistance directly to some enterprises with high pollution and high dependence on resources, it enables local enterprises to have more capital and opportunities to seek transformation, promotes the incubation of new industries through green technology innovation and other means, and optimizes the original industrial structure of cities. It also helps improve the utilization efficiency of resources and reduce their dependence on local resources. Therefore, the transfer payment policy can affect the resource dependence of resource-depleted cities through industrial transformation. On the other hand, reducing the degree of resource dependence also means protecting local non-renewable resources, protecting the natural environment, and reducing the emission of pollutants. Especially in the context of the national vigorous implementation of air governance, the transfer payment policy provides support and motivation for these resource-depleted cities. This ensures environmental benefits and economic benefits. At the same time, this study makes a corresponding contribution to the research on the impact of transfer payment policies on local resource dependence.

Firstly, our study supplements the existing literature on the impact of transfer payment policies on urban resource dependence and the relationship between transfer payment policies and regional development [45,46,47,48,49]. We reviewed the existing research methods [15,59], reconstructed the model for research, and selected ten control variables (such as lngdp, lngov, and lnden [27,28,29,30,31,32,33,34,35,61,62]) for testing and obtained the results. We emphasize that there is an inverted U-shaped relationship between the degree of resource dependence and economic development, and transfer payment policies significantly reduce the degree of resource dependence of resource-depleted cities. In addition, a series of analyses, including a baseline regression test, placebo test, elimination of variable bias test, and interference policy impact test, were carried out to enrich our study, indicating the robustness of the baseline regression results. The implementation of the transfer payment policy can still significantly reduce the degree of resource dependence of resource-depleted cities. This helps further improve our understanding of the relationship between resource dependence and economic development. This test will guide similar follow-up studies.

Secondly, the study added the analysis of influencing factors and excluding interfering factors for regional differences in dependency degree and revealed the correlation between dependency degree and variables such as regional enterprise structural adjustment and technological innovation [54,55]. The results indicate that the transfer payment policy has a more obvious effect on the central and western regions of China, while the impact on the resource dependence of the eastern region is not significant. It is helpful for each region of China to adopt different financial measures according to local factual conditions to reduce the dependence on the resources of cities and promote sustainable economic development in each region. It enriches the theory of regional development and is of great significance to the regional coordinated development of the regional economy.

Finally, the study has a certain reference significance for the transfer payment policy to promote the transformation of industrial structure. Through mechanism analysis, it is found that the resource dependence of resource-depleted cities can be reduced through the rationalization of industrial structures, the upgrading of industrial structures, and the enhancement of infrastructure construction and technological progress. This puts forward new requirements for the transformation of industrial structures in resource-depleted cities. Under the influence of transfer payment policies, industrial structure transformation can help cities to reduce their dependence on resources, promote the sustainable development of resources, and then feed the sustainable development of society. Therefore, this study can provide new guidance for sustainable development.

6. Conclusions

This study is an empirical study of the impact of transfer payment policies on resource-depleted cities using the difference-in-differences model based on annual data from 113 resource-based cities in China from 2006 to 2017. This study is not directly identified by a special theory, but it is an empirical one. Therefore, through the experimental evaluation of the effect and mechanism, the study draws the following conclusions about the correlation between transfer payment policy and resource dependence of resource-depleted cities: Firstly, transfer payment policy significantly reduces the degree of resource dependence of resource-depleted cities, and the degree of resource dependence presents an inverted U-shaped relationship with economic development. The expansion of the number of industrial enterprises also helps to reduce the degree of resource dependence to some extent. Secondly, numerous analyses show that the baseline regression results are robust. Thirdly, through heterogeneity analysis, it is found that due to the heterogeneity of economic development, transfer payment policies have more obvious effects on the central and western regions of China, while the impact on resource dependence in the eastern region is not significant. In addition, mechanism analysis shows that the transfer payment policy has an impact on the promotion of the transformation of industrial structures through the rationalization of industrial structures, the upgrading of industrial structures, and the enhancement of infrastructure construction and technological progress.

Based on the above findings, this study puts forward the following five policy proposals:

Firstly, the government should continue to improve the financial policies provided to resource-depleted cities, continuously enhance the positive effects of financial policies on resource-depleted cities by reducing resource dependence, and design a more systematic method of capital utilization to balance the contradiction between industrial transformation and social stability. Secondly, the government should increase its support to the resource-depleted cities in the central and western regions. Although the central and western regions have weak basic conditions for their own development due to their geographical location, the transfer payment policy has positive effects on these regions. Therefore, it is necessary to continue to expand the policy effect, establish a more stable long-term mechanism, and accelerate the economic development of the central and western regions. Thirdly, re-exploring the way to reduce resource dependence in the eastern region is of great importance. The economic development of the eastern region is rapid, but transfer payments cannot provide the impetus to reduce local resource dependence. It is necessary to find a more appropriate assistance model for the eastern region to help improve other cities. The problem of resource depletion is what many resource-based cities face and will face. The transformation mode and path of resource-depleted cities are worthy of other resource-based cities to learn and imitate. Moreover, the central government should summarize the individual problems existing in the transformation of resource-depleted cities in different regions, so as to avoid the same problems from recurring in resource-based cities. Last but not least, the local government should continue to expand the intensity of industrial transformation, promote the evolution of the secondary industry to the tertiary industry, continue to carry out green technology innovation, introduce high-quality talents, improve the efficiency of resource utilization, and reduce urban pollution. Resource-depleted cities can get rid of the “curse of resources” and enter a benign development mode.