Advertisement

Environmental Science and Pollution Research

, Volume 26, Issue 12, pp 12136–12149 | Cite as

Study on the spatial correlation structure and synergistic governance development of the haze emission in China

  • Hao LiEmail author
  • Ming ZhangEmail author
  • Chen Li
  • Man Li
Research Article
  • 121 Downloads

Abstract

To clarify the current situation of haze emission and governance in China, the study analyzed the characteristics of spatial correlation structure and synergistic governance development of the haze emission of 31 provinces in China, based on social network analysis and distance synergistic model. The results indicated that the spatial correlation of inter-provincial haze emission in China presented a typical “central–marginal” network structure. The provinces in the network center were mostly located in the Beijing–Tianjin–Hebei region and the Yangtze River Delta region. The synergistic governance development of haze in China showed a lower level and fluctuating upward trend. In addition, the increase of network density, the decline of network grade, and the decrease of network efficiency would all improve the level of synergistic governance development. Therefore, focusing on the haze of the central provinces, improving the network structure, and improving regional synergy are important measures for effective governance. This paper improves the previous research model, considers the impact of economic and demographic factors on haze pollution, establishes a new model for analyzing spatial correlation structure of haze and calculating the synergistic governance level of haze, and designs feasible ways to raise the synergistic governance level of haze in China.

Keywords

Haze emission Spatial correlation Synergistic governance development 

Introduction

With the development of industrialization and urbanization of all countries in the world, the problem of haze pollution continues to grow, restricting the sustainable development of society and the global economy (Lin and Wang 2016; Tong et al. 2016), even posing a great threat to human health (Andersen et al. 2012; Hart et al. 2015; Chen et al. 2017). As the second largest economy in the world, China’s economy has been in the stage with medium–high growth after maintaining rapid growth for nearly 30 years. However, China is still one of the countries with the most serious haze pollution in the world, yet haze prevention and control work is facing tremendous pressure. Meanwhile, due to the spatial propagation characteristics of haze, the haze emissions among regions are not independent, but present a certain spatial correlation. Therefore, independent governance has little effect. In this context, defining the spatial relationship of the haze emission in China and achieving the regional synergistic governance of haze pollution are the current research hotspots.

In terms of spatial correlation, taking the measure of spatial correlation as the goal, Ding et al. (2017) adopted exploratory spatial data analysis (ESDA) to analyze the spatial correlation characteristics of haze pollution. The results demonstrated that there were obvious spatial agglomeration features. Li et al. (2018a) evaluated the gravity center revolution of air pollution in Yangtze River Delta of China based on gravity model and discovered that since 2014, the gravity center of air pollution in Yangtze River Delta continued westward with the characteristic of the north–south circular movement. Li et al. (2018b) employed the spatial Durbin model (SDM) to explore the spatial spillover effects of industrialization and urbanization factors on haze emission in China’s Huang–Huai–Hai region. Liu and Du (2018) identified the spatial dependence relationship of haze pollution among 279 cities in China based on the nonparametric approach of Granger causality. Filonchyk et al. (2018) used linear correlation analysis to analyze the spatial correlation among multiple pollutants in the city of Lanzhou, China. Zhang et al. (2018) built a spatial indicator that measured the impact of near areas’ pollution to account for the spatial correlation of pollution by spatial econometric measurement methods. Most of the abovementioned papers did not consider relevant factors that affect haze emissions.

From a social perspective, more social factors related to air pollution, such as economic and demographic factors, should be fully taken into account. Economic development is an important cause of serious haze pollution, and people are the main actors of economic activities. Liu and Pei (2017) conducted an empirical study on the environmental Kuznets curve hypothesis (EKC), using haze pollution data from 160 Chinese cities, and found that economic level and population scale had significant positive effects on haze pollution in China. Therefore, “economic level” and “population scale” are important factors affecting air pollution. Because of relating to economic and demographic factors, traditional spatial measurement methods tend to limit spatial correlation to geographical “adjacent” or “close” regions (Li et al. 2014a) when describing spatial correlation structure of haze. Influenced by geographical and meteorological factors, haze pollution mainly diffuses to adjacent or close areas. However, the focus of China’s economic development is multi-directional cooperation among coastal provinces and inland provinces, and with the rapid development of transportation industry, regions with “economic” or “population” relationship may not be geographically adjacent or close. In addition, taken into account the three factors of “haze emission,” “economic level,” and “population scale,” the spatial correlation structure between regions becomes more complex and multithreaded, showing complex network structure characteristics. The existing literature does not consider the spatial correlation characteristics of haze pollution in China under the influence of economic and demographic factors. Therefore, this paper attempts to use the “social network analysis” to remedy this shortcoming. Social network analysis can break through geographical restrictions and analyze regional spatial correlation characteristics from the overall view.

In the realm of synergistic governance of haze, Wang (2014) analyzed the phenomenon of air quality improvement in Beijing during the 2014 APEC meeting and found that synergistic air pollution control could achieve significant effects. Xie et al. (2016) designed a synergistic emission reduction model to reduce the harmful effects of air pollution on the Yangtze River Delta and to reduce the pollutant discharge fee. A study by Lee (2017) discovered the Japanese government had made very limited achievements in haze pollution control in recent years and further clarified the necessity of synergistic governance of haze among China, Japan, and South Korea. Carpenter (2018) analyzed the oil pollution entering the marine environment and found international cooperation was an important way to improve marine environment. Through evolutionary game research, Zhang and Li (2018) demonstrated local governments could not spontaneously form the synergistic model and required the superior government to implement supervision. The current research on synergistic governance of haze is mostly focused on the necessity of cooperation, system design and countermeasures, but less on the overall analysis or evaluation of synergistic level of haze governance. Through the description and analysis of the synergistic governance level of haze, the changing trends and existing problems of regional environmental cooperation can be found, which is of great value to the formulation of environmental policies. In addition, when calculating the synergistic governance level of haze, economic level and population size are also factors that cannot be ignored. The improvement of economic level and the expansion of population size are important indicators of social development, while the improvement of environmental quality is also an embodiment of social development. In this context, this paper comprehensively considers the three factors of “haze,” “economic,” and “population” in the region and is devoted to calculating the level of social development in China on the basis of “synergistic governance of haze,” which is named as “synergistic governance development of the haze emission in China.” On this basis, this paper further analyzes the effect of the spatial correlation structure on synergistic governance development.

This paper attempts to make up for the gap of existing research in fields of the spatial correlation structure of haze emission and the level of Chinese social development based on the synergistic governance of haze in the context of comprehensive consideration of regional haze, economy and population. It also makes an in-depth analysis of the core areas and key factors affecting haze pollution. The main contributions and innovations of this paper are as follows. First of all, two important factors, economic level and population scale, which affect haze pollution, are added into the analysis model to enhance its explanatory ability and expand the research contents in the field of haze pollution. Secondly, this paper improves the previous research model and establishes a new model to analyze the spatial correlation structure of haze pollution so as to calculate the synergistic governance level of haze. Thirdly, based on the research results, this paper designs feasible ways to enhance China’s capability in terms of regional synergistic haze control.

The rest of the paper is organized as follows: the second section introduces the methodologies and data involved in this paper; the third section describes the research results and related discussions; and the fourth section summarizes the paper and puts forward the policy recommendations.

Methodologies and data

Construction of spatial correlation structure of the haze emission

Social network analysis (SNA) is a quantitative analysis method developed by sociologists based on mathematical methods and graph theory. Through the analysis of the correlation in the network, it takes “relationship” as the basic analysis unit to discuss the structure and attribute characteristics of the network (Borgatti et al. 2009). Social network analysis can break through geographical restrictions, analyze regional spatial correlation characteristics from the overall view, quantitatively analyze the individual role played by each region in the whole network, locate the corresponding areas of key nodes in the network, and fully mobilize and guide the role of key areas. Based on the analysis results, the regulatory path to promote the development of the correlation network in a favorable direction could be found.

Schiavo (2010) and Cassi et al. (2012) employed the social network analysis method to study the network characteristics of international trade and financial integration and testified that most international financial transactions and industrial cooperation were carried out through a few countries as the central nodes. Guo (2014) analyzed the spatial correlation network of carbon emissions in China’s industrial sector, discovered that the industrial sector was at the core of the carbon emission network, and proposed to implement specific energy saving and emission reduction measures using the linkage effect generated by the core sector. Sun et al. (2016a) analyzed the spatial correlation network of inter-provincial carbon emission in China and found that the eastern coastal cities in China were located at the core of the carbon emission network, which was easy to affect other provinces. Li et al. (2018) explored the key factors and mechanisms affecting the service dynamics of green construction enterprises. The above literature used social network analysis to study the spatial association characteristics of multiple subjects and explore the core influencing factors.

The key to social network analysis is the establishment of “relationship” (Waters 2014). “Relationship” refers to the correlation among behavior subjects. The current portrayal of spatial correlation is mainly based on the VAR Granger causality test (Groenewold et al. 2010) and gravity model (Hou et al. 2009). Considering that the network rested on the VAR model cannot describe the trend of spatial evolution, the gravity model is not only applicable to the total data but also beneficial to take economic geography factors into consideration comprehensively.

The gravity model is a method to measure the spatial correlation effect, which can quantitatively calculate the attraction between regions. The basic assumption is that the spatial correlation between regions has a law of mutual attraction and that the attraction is directly proportional to the spatial attribute index. Meanwhile, according to the principle of distance attenuation (Yang 1989), the attraction decreases with the increase of distance between regions. So the spatial correlation intensity between regions can be calculated.

The gravity model is derived from the law of universal gravitation proposed by British physicist Isaac Newton in the seventeenth century (Newton Sir et al. 2007).
$$ F=G\frac{M_1{M}_2}{R^2} $$
(1)

Here, F represents the gravitational force between any two objects; M1 and M2 stand for the mass of two objects respectively; R denotes the distance between the two objects; and G is the gravitational constant. Formula (1) indicates that the gravitational force between any two objects is proportional to their own masses and inversely proportional to the square of the distance between them.

Ravenstein, a British scholar, first introduced the gravity model into the field of social science research for the analysis of population migration (Ravenstein 1885). Reilly, an American scholar first applied the Gravity Model to the study of economics (Reilly 1929). He proposed the famous law of retail gravity, considering that among the commodities retailed in the two regions, the retail sales of goods in the same region are directly proportional to the population of the retail site and inversely proportional to the square of the distance from the retail site to the source of supply. Later, the gravity model has been widely applied in economic research. The American linguist Zipf deemed that the correlation between regions was directly proportional to the population within the region and inversely proportional to the distance between the two regions (Zipf 1946). Regional economist Isard also put forward the same view (Isard 1956). Tinbergen and Poyhonen were the first two scholars to apply the gravity model to international trade research (Tinbergen 1962; Poyhonen 1963). They argued that the individual trade flows between two countries were proportional to the size of their respective economies and inversely proportional to the distance between them. The earliest scholars in China who employed the gravity model for economic research were Wang and Zhuang (Wang and Zhuang 1996). They two conducted a quantitative analysis of the economic contact between two Chinese regions, SuXiChang and Shanghai.

In the early economic studies, the Gravitational Model was applied as an empirical model based on practice and logic. Its general empirical form was:
$$ Y=K\cdotp f\left({x}_1,{x}_{2,}\cdots, {x}_n\right) $$
(2)

Among them, xi represents relevant metrics (such as population, output value, spatial distance) and K is the gravitational coefficient.

Based on the generalized fractal hypothesis, Chinese scholars Chen and Liu proved the power function form of the gravity model, which made it develop from an empirical model to a theoretical model (Chen and Liu 2002), as shown in Formula (3):
$$ Y=K\frac{M_i^{\alpha }{M}_j^{\beta }}{D^{-\gamma }} $$
(3)

Here, Mi andMj represent attribute index of region i and region j respectively; D is the distance between the two regions; K is the gravitational coefficient; α and β are the index dimension coefficients, and−γis the distance attenuation coefficient.

According to the allometric growth relationship (Bertalanffy 1969), the generalized dimension of the region tends to be consistent theoretically. Through empirical research, Taaffe found the spatial correlation strength between regions is inversely proportional to the square of the distance (Taaffe 1962), that is, the distance attenuation coefficient is set to 2.

Formula (3) can be converted to:
$$ Y=K\frac{M_i{M}_j}{D^2} $$
(4)

Formula (4) is the general theoretical form of the gravity model applied in economic research (Head 2003).

In this paper, the gravity model is used to describe the spatial correlation of the haze emission in China. First, the general form of the gravity model needs to be improved.

On the one hand, when the subject is an object, the attractive forces are mutual and equal, but when the subject is a specific area in economic research, in view of the obvious differences in economic, technological, and social factors between two regions, the economic contact between two regions shows a certain unidirectionality, and the attraction of the two regions to each other cannot be exactly the same. The gravitational coefficient K needs to be reasonably set to show the differences. Through empirical research, Wang et al. (2006) held that the gravitational coefficient K should indicate the proportion structure corresponding to the research topic. In the study of regional economic integration, Hou et al. (2009) set the gravitational coefficient K as the proportion of GDP in a certain region to the total GDP. Sun et al. (2016b) set the gravitational coefficientKas the proportion of carbon emission in a certain region to total carbon emission when studying the spatial correlation of provincial carbon emission in China. Li and Niu (2017) set the gravitational coefficientKas the proportion of innovation output in a certain region to the total output when studying the spatial correlation of regional innovation output.

In this paper, the gravity coefficient K is set as the proportion of haze emission in one region to the total emission, so as to calculate the gravity of haze emission in this region to the other region.

On the other hand, for the measurement of distance, the spatial distance between the center of gravity of two objects is used in physics. But in the study of the spatial correlation of haze emission, in addition to the geographical difference, the economic difference is also an important factor affecting the regional gravity. And the littler the geographical difference, the greater the economic difference and the greater the regional gravity will be. Referring to Liu’s treatment of “distance” in the study of spatial correlation of energy consumption in China (Liu et al. 2015), this paper calculates the value that the spatial distance between the two regions is divided by per capita GDP difference, to describe the comprehensive impacts of geographical and economic differences on spatial correlation of haze emission.

In addition, “economic level” and “population scale” should also be important indicators for characterizing regional attributes. At the same time, the influence of these three indexes on spatial correlation is a nonlinear action, which can be represented by geometric mean value (Liu and Dai 2013; Tang et al. 2013). So this paper uses the geometric average values of “haze emission,” “economic level,” and “population size” as the regional attribute indexes.

To sum up, the gravitational model constructed in this paper is as follows:
$$ {Y}_{ij}=\frac{P_i}{P_i+{P}_j}\times \frac{\sqrt[3]{R_i{P}_i{G}_i}\times \sqrt[3]{R_j{P}_j{G}_j}}{{\left(\frac{D_{ij}}{g_i-{g}_j}\right)}^2} $$
(5)

In Formula (5), Yij represents the gravity of haze emission between region i and region j; Pi and Pj stand for the total amount of haze emissions in region i and region j respectively; Ri and Rj indicate the total population of region i and region j; Gi and Gj denote regional GDP of region i and region j respectively; Dij refers to the spatial distance between the central city of region i and region j, and gi and gj are GDP per capital of region i and region j respectively.

According to Formula (5), the gravity of provincial haze emission can be calculated. However, the spatial gravity at this time is an attribute datum and cannot represent the relationship. So on this basis, the spatial correlation must be further defined, that is, the gravity matrix should be converted to the spatial correlation matrix. This article refers to the “threshold selection method” adopted by Wu et al. (2017), Zhang and Lu (2017), Li et al. (2018), and others: if the element is larger than the average value of its row, the conversion is 1, indicating that the provinces in the row and the provinces in the column have a haze emission correlation; if the element is smaller than the average value of its row, the conversion is 0, demonstrating that the provinces in the row and the provinces in the column have no haze emission correlation.

Overall network structure characteristics index of spatial correlation network

Overall network structure characteristics are usually qualified by density, connectedness, grade, and efficiency.

Density reflects the closeness of spatial correlation network: the greater density, the more closely correlation among haze emissions, and the greater impact of network structure on the haze emissions of each province will be. The formula for calculating density is:
$$ D=\frac{M}{N\times \left(N-1\right)} $$
(6)

In Formula (6), N is the number of provincial correlations in the network. N × (N − 1) represents the maximum number of possible correlations in the network, while M is the number of correlations actually existing in the network.

Connectedness reflects the robustness and vulnerability of spatial correlation network. The greater the degree of network correlation, the existence of the direct or indirect correlation among cities in the network can be demonstrated. The formula for calculating connectedness is:
$$ C=1-2\times \frac{U}{N\times \left(N-1\right)} $$
(7)

In Formula (7), U is the logarithm of the number of unreachable points in the network.

Grade reflects the hierarchical structure of the spatial correlation network. The higher the grade, the more provinces will be subordinate and marginal in the network. The formula for calculating grade is:
$$ H=1-\frac{S}{\max (S)} $$
(8)

In Formula (8), S is the logarithm of the symmetric reachable points in the network, and max(S) is the logarithm of the maximum number of symmetrically reachable points in the network.

Efficiency shows the correlation efficiency of the spatial correlation network. The lower the network efficiency, the more connections among provinces in the network, and the more spatial spillover channels will be. And it will be easier to promote the space flow of haze through spatial correlation network to narrow the comparative advantages among provinces. The formula for calculating efficiency is:
$$ E=1-\frac{O}{\max (O)} $$
(9)

In Formula (9), O is the number of redundant lines in the network, max(O) is the largest number of redundant lines in the network.

Individual network structure characteristics index of spatial correlation network

Individual network structure characteristics are normally described by three indexes as degree centrality, betweenness centrality, and closeness centrality.

Degree centrality reflects the extent to which Chinese provinces are centrally located in the spatial correlation network of the haze emissions. The greater the degree centrality, the closer the province is to the center of the network in the network, and the more connections it has with other provinces. The formula for calculating degree centrality is:
$$ {C}_d(i)=\frac{d(i)}{N-1} $$
(10)

In Formula (10), d(i) is the number of direct relationships between province i and other provinces in the network.

Betweenness centrality shows the degree to which a province controls other provinces in the spatial correlation network. The greater the betweenness centrality, the stronger the “mediator” role the province has. The formula for calculating betweenness centrality is:
$$ {C}_b(i)=\frac{2\sum \limits_{j<k}\left[\frac{l_{jk}(i)}{l_{jk}}\right]}{\left(N-1\right)\left(N-2\right)} $$
(11)

In Formula (11), ljk is the number of relationship paths existing between province j and province k; ljk(i) is the number of lines that pass through province i on the relationship paths between province j and province k.

Closeness centrality reflects the extent to which a province in spatial correlation network is not controlled by other provinces. The higher the closeness centrality, the more direct correlations between this province and other provinces will be. The province is the central actor in the network. The calculation formula of closeness centrality is:
$$ {C}_c(i)=\frac{1}{\sum \limits_{j=1}^n{q}_{ij}} $$
(12)

In Formula (12), qij is the shortcut distance between province j and province k (i.e., the number of lines contained in the shortcut distance).

Specifically, the spatial correlation network reflects the relational data, so the calculation formula should adopt the orientation form, but the structural index reflects the attribute data, so the calculation formula should adopt the non-directional form.

Measure of synergistic governance development level of the haze emission

Due to the existence of externalities, synergistic governance has become an inevitable trend to solve the problem of haze emission. In recent years, research on synergistic governance development level of the haze emission in China has been a hotspot. In this paper, the “distance synergistic model” is used to measure synergistic governance development of the haze emission in China.

Based on synergetic as well as TOPSIS method, “distance synergistic model” defines the order parameters and order degree of the system. First, by measuring the distance between the order degree of the existing state of the subsystem and its ideal state or standard, to judge whether the development of each order parameter of the subsystem is orderly and synergistic, and then to calculate synergistic development level of the composite system as a whole (Zheng, 2010).

Based on the index system adopted in this paper, in order to ensure the consistency of the study, three indexes, “GDP,” “population scale,” and “haze emissions,” have been selected to build the “distance synergistic model”. The specific steps are as follows:
  1. 1.

    Normalization of indexes

     
For high-value indexes (such as GDP and population scale) with larger values and better attributes, the formula can be converted as:
$$ {a}_{ijt}=\frac{\left({x}_{ijt}-\underset{t}{\min }{x}_{ijt}\right)}{\left(\underset{t}{\max }{x}_{ijt}-\underset{t}{\min }{x}_{ijt}\right)} $$
(13)
For low-value indexes with lower values and better attributes (such as haze emission), the formula can be converted as:
$$ {a}_{ijt}=\frac{\left(\underset{t}{\max }{x}_{ijt}-{x}_{ijt}\right)}{\left(\underset{t}{\max }{x}_{ijt}-\underset{t}{\min }{x}_{ijt}\right)} $$
(14)
where xijt remarks the jth index of the subsystem i in period t, and aijt represents the normalized index.
  1. 2.

    Measure of the distance between the subsystem’s existing state and its ideal state

     
After the normalization of each index, the maximum value 1 is taken as the positive ideal point, the minimum value 0 as the negative ideal point, and the distances between each subsystem and its positive ideal points and negative ideal points, that is, \( {D}_{it}^{+} \)and\( {D}_{it}^{-} \) respectively, are calculated. The calculation formula is:
$$ \Big\{{\displaystyle \begin{array}{l}{D}_{it}^{+}=\sqrt{\sum \limits_j{\left(1-{a}_{ijt}\right)}^2}\\ {}{D}_{it}^{-}=\sqrt{\sum \limits_j{\left({a}_{ijt}\right)}^2}\end{array}} $$
(15)
  1. 3.

    Measure of development degree of subsystem dit

     
$$ {d}_{it}=\frac{D_{it}^{-}}{\left({D}_{it}^{-}+{D}_{it}^{+}\right)} $$
(16)
dit is between 0 and 1. It indicates the relative development degree of the subsystem i and its ideal state. The larger the value, the higher the development degree of subsystem will be.
  1. 4.

    Measure of development degree of composite system dt

     
$$ {d}_t=\sqrt[k]{\prod \limits_i^k{d}_{it}} $$
(17)
  1. 5.

    Measure of the ideal development degree of subsystem dit

     
$$ {d_{it}}^{\prime }=\sum \limits_j^k{\omega}_j{\alpha}_{ij}{d}_{jt} $$
(18)

The ideal development degree of a subsystem indicates the development degree it should achieve under the associated impact of all subsystems in the composite system. ωj is the weight of the subsystem j. The 31 regions studied in this paper are geographical units of the same administrative level in China, so ωj = 1/31. αij is the pull factor of the subsystem j to the subsystem i, and the pull factor of the subsystem to itself is 1. In line with the research of Li et al. (2014b) and Li et al. (2017), the pull factor in this paper is calculated by gray relational analysis.

Gray relational analysis measures the degree of correlation between subjects based on the degree of similarity or dissimilarity between the developments of the subjects. It reveals the characteristics and degree of dynamic correlation of things. If the relative changes of the subject in the development process are basically the same, the degree of correlation between the two is considered to be large; on the contrary, the degree of correlation is small. The technical details are as follows:

Assume there is a set of reference series as follows:
$$ Y=\left({y}_1,{y}_2,{y}_3,\cdots, {y}_k,\cdots, {y}_n\right). $$
There are m groups of comparisons as follows:
$$ {X}_i=\left({x}_{i1},{x}_{i2},{x}_{i3},\cdots, {x}_{ik},\cdots, {x}_{in}\right),\kern0.36em i=1,2,3,\cdots, m. $$
The correlation coefficients between points in sequence Xi and points in sequence Y are defined as follows:
$$ {\xi}_{ik}=\frac{\underset{i}{\min}\underset{k}{\min}\mid {x}_{ik}-{y}_k\mid +\rho \underset{i}{\max}\underset{k}{\max}\mid {x}_{ik}-{y}_k\mid }{\mid {x}_{ik}-{y}_k\mid +\rho \underset{i}{\max}\underset{k}{\max}\mid {x}_{ik}-{y}_k\mid } $$
(19)

In the formula, ρ is the resolution coefficient, and the value is taken in (0,1). The smaller ρ, the greater the difference of correlation coefficients between subjects, and the stronger the discrimination ability is. Usually ρ is 0.5.

The correlation degree of sequence Xi to column Y is defined as follows:
$$ {\gamma}_i=\frac{\sum \limits_{k=1}^n{\xi}_{ik}}{n} $$
(20)
When calculating the correlation degree among multiple sequences, the correlation degree matrix R can be constructed.
$$ R=\left[\begin{array}{cccc}{\gamma}_{11}& {\gamma}_{12}& \cdots & {\gamma}_{1\mathrm{n}}\\ {}{\gamma}_{21}& {\gamma}_{22}& & {\gamma}_{2n}\\ {}\vdots & \vdots & & \vdots \\ {}{\gamma}_{m1}& {\gamma}_{m2}& \cdots & {\gamma}_{mn}\end{array}\right] $$
(21)
γij stands for the degree of correlation between subject j and subject i. This paper employs the degree of correlation γij to represent the pull factor αij.
  1. 6.

    Measure of synergistic degree of the subsystem cit

     
$$ {c}_{it}=\frac{d_{it}}{d_{it}+\mid {d}_{it}-{d_{it}}^{\prime}\mid } $$
(22)
cit is between 0 and 1, indicating the actual synergistic degree of the subsystem i, compared with the ideal development degree under the associated impact of all subsystems. The greater the value, the higher the synergistic degree of the subsystem.
  1. 7.

    Measure of synergistic degree of the composite system ct

     
$$ {\mathrm{c}}_t=\sqrt[k]{\prod \limits_i^k{c}_{it}} $$
(23)
  1. 8.

    Measure of synergistic development level of the composite system CDt

     
$$ {\mathrm{CD}}_t=\sqrt{c_t{d}_t} $$
(24)

Data sources

In this paper, 31 Chinese provinces (excluding Hong Kong, Macao, and Taiwan) are taken as network nodes to empirically investigate the spatial correlation of haze emissions. The period of the sample is from 1998 to 2016. The sources of data needed for gravity model calculation are as follows: the provincial GDP, population, and per capita GDP are derived from the corresponding year’s China Statistical Yearbook. The geographical distance between provinces is represented by the spherical distance between the central cities and calculated by ArcGIS.

Data on haze emission of all provinces are collected from the Global Fine Particulate Matter (PM2.5) Concentrations Inventory (Van et al. 2016), collected by CIESIN at Columbia University, and the data are collected by the end of 2016. This is a global high-resolution list based on satellite monitoring, regardless of the possibility that satellite monitoring might be affected by meteorological factors so its accuracy may be slightly lower than ground monitoring. However, the spatial distribution of PM2.5 concentrations in the same city can also be different, and the ground monitoring can only approximately estimate the PM2.5 concentrations in the whole region based on point source data. On the contrary, the satellite monitoring data are non-point source data, which can fully reflect the PM2.5 concentrations and its changing trend in the whole region. It has become an important reference for the global atmospheric chemical simulation and the compilation of atmospheric pollution inventory (Shao et al. 2016). Therefore, the global average annual PM2.5 concentration data are sufficient for haze emission study conducted in this paper.

The concentration data can be downloaded for free,1 while the obtained file is not a direct data table, but in Network Common Data format. This article employees ArcGIS 10.2 software to draw PM2.5 concentrations data (μg/m3) from NC files. The specific steps are as follows: first, the vector data and PM2.5 raster data of 31 Chinese provinces (municipalities) are superimposed; second, the PM2.5 annual haze emission of each province (municipality) is counted with regional statistical tools.

Results and discussion

Analysis of characteristics of spatial correlation structure of haze emission

According to the gravity model shown as Formula (5), this paper firstly calculates the gravity matrix of inter-provincial haze emission in China and draws the spatial correlation structure of haze emission in 2016 by using the visualization tool NetDraw in UCINET 6.0 software. The specific results are exhibited in Fig. 1.
Fig. 1

Spatial correlation structure of haze emission in China in 2016

It can be seen from Fig. 1 that the spatial correlation of inter-provincial haze emission in China displays a typical “central–marginal” network structure, with Beijing, Tianjin, Shanghai, Jiangsu, and Zhejiang at the center of network, and the remaining provinces basically on the edge. The whole network is divided into center area and edge area. Among them, there are 230 correlations among 31 provinces, and each province has multiple spatial correlations, which indicates that the haze emission in China is generally correlated in space and that there is a stable spatial correlation structure of haze emission among the provinces.

When province i produces gravitational effects on province j, province i is the gravitational city of province j, and province j is the reverse-gravitational city of province i. The more reverse-gravitational cities a province has, the stronger its gravitational effect will be; the more gravitational cities a province has, the stronger its reverse-gravitational effect will be. The gravity strength of China’s provinces in 2016 is revealed in Fig. 2, and the reverse-gravity strength of China’s provinces in 2016 is shown in Fig. 3.
Fig. 2

The gravity strength of China’s provinces in 2016

Fig. 3

The reverse-gravity strength of China’s provinces in 2016

From Fig. 2, Chinese eastern coastal provinces have higher gravity strength, while the gravity strength of the western inland provinces (such as Xinjiang and Tibet) is 0, and other provinces have medium gravity strength. It can be seen from Fig. 3 that the reverse-gravity strength of the western provinces of China and the southeastern coastal provinces is higher, while the reverse-gravity strength of Hebei, Jiangsu, and Anhui provinces is the lowest, and other provinces’ is at a medium level.

According to Formulas (6) to (9), the network density, network correlation degree, network grade, and network efficiency are calculated respectively to analyze the overall characteristics of spatial correlation structure of haze emission from 1998 to 2016 in China. The results are shown in Fig. 4.
Fig. 4

The overall characteristics of spatial correlation structure of haze emission in China

As shown in Fig. 4, from 1998 to 2016, the network density of spatial correlation structure of inter-provincial haze emission in China shows an upward trend in general, indicating that the inter-provincial correlation of haze emission in China is strengthening. The correlation degree of the network is 1, which indicates that the network structure is stable and there is a universal spatial spillover and correlation effect. In addition, the network grade and network efficiency of the spatial correlation network demonstrates a downward trend in general, showing that the formerly relatively strict spatial correlation structure of haze emission has been gradually broken, that the correlation and mutual influence between inter-provincial haze emission have strengthened, and that the stability of the network has gradually improved.

According to the Formulas (10) to (12), the point centrality, betweenness centrality, and closeness centrality are calculated respectively to analyze the individual structure characteristics of the spatial correlation network of haze emission in China in 2016. The results are shown in Table 1.
Table 1

Individual centrality of spatial correlation network of the haze emissions in Chinese provinces

Province

Point centrality

Betweenness centrality

Closeness centrality

Figure

Ranking

Figure

Ranking

Figure

Ranking

Anhui

26.667

18

0.330

19

57.692

18

Beijing

76.667

3

10.379

3

81.081

3

Chongqing

26.667

18

0.294

21

57.692

18

Fujian

46.667

8

2.179

8

65.217

8

Gansu

40.000

9

0.958

10

62.500

9

Guangdong

50.000

7

3.091

6

66.667

7

Guangxi

23.333

25

0.241

23

56.604

25

Guizhou

26.667

18

0.396

18

57.692

18

Hainan

20.000

28

0.138

28

54.545

31

Hebei

20.000

28

0.129

29

55.556

28

Henan

30.000

14

0.494

15

58.824

14

Heilongjiang

23.333

25

0.152

27

56.604

25

Hubei

30.000

14

0.601

13

58.824

14

Hunan

26.667

18

0.298

20

57.692

18

Jilin

20.000

28

0.089

31

55.556

28

Jiangsu

86.667

2

11.097

2

88.235

2

Jiangxi

23.333

25

0.195

26

56.604

25

Liaoning

20.000

28

0.106

30

55.556

28

Inner Mongolia

40.000

9

1.325

9

62.500

10

Ningxia

26.667

18

0.198

24

57.692

18

Qinghai

26.667

18

0.264

22

57.692

18

Shandong

56.667

6

2.806

7

69.767

6

Shanxi

26.667

18

0.198

24

57.692

18

Shaanxi

33.333

11

0.470

17

60.000

11

Shanghai

90.000

1

12.418

1

90.909

1

Sichuan

33.333

11

0.740

11

60.000

11

Xinjiang

30.000

14

0.740

11

60.000

11

Tianjin

70.000

5

7.767

4

75.000

5

Tibet

33.333

11

0.740

11

60.000

11

Yunnan

30.000

14

0.500

14

58.824

14

Zhejiang

73.333

4

7.119

5

78.947

4

From Table 1, it can be seen that the point centrality, betweenness centrality, and closeness centrality of Beijing, Tianjin, Jiangsu, Shanghai, and Zhejiang are all in the top five and significantly larger than other provinces. It suggests that the five regions are centrally located in the spatial correlation structure of haze emission and have more connection with other provinces and play an important role of “mediator” and “bridge” to connect with other provinces more quickly. In other words, these five regions above play a role of central actor in the network. This is mainly because the abovementioned provinces are located in China’s Beijing–Tianjin–Hebei and the Yangtze River Delta economic zone. They have absolute advantages in economic development, industrial structure, energy consumption and other aspects, so they have a huge leading and demonstrative role in the synergistic governance of environmental pollution across the country. Therefore, fully exerting the synergistic leading role of Beijing–Tianjin–Hebei region and Yangtze River Delta region will be the key to further promoting the synergistic governance of haze emission in China.

Analysis of synergistic governance development of haze emission

According to the distance synergistic model described by Formulas (13) to (21), the synergistic governance development of inter-provincial haze emission in China from 1998 to 2016 has been calculated. The results are shown in Table 2.
Table 2

The synergistic governance development of inter-provincial haze emission in China

Year

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

Synergistic index

0.589

0.581

0.584

0.593

0.584

0.582

0.582

0.583

0.592

0.582

Year

2008

2009

2010

2011

2012

2013

2014

2015

2016

Mean

Synergistic index

0.588

0.593

0.586

0.594

0.590

0.600

0.593

0.591

0.595

0.589

Its annual changing trend is displayed in Fig. 5.
Fig. 5

The annual changing trend of the synergistic governance development of inter-provincial haze emission in China

It can be seen from Fig. 5 that during the study period, the synergistic governance development level of inter-provincial haze emission in China shows a fluctuating upward trend, and the fluctuation amplitude is large. The synergistic governance development level of inter-provincial haze emission has increased from 0.589 in 1998 to 0.595 in 2016, with an overall increase of only 0.006. It suggests that the synergistic governance development level of inter-provincial haze emission is not high, the synergistic state is unstable, its overall growth rate is slow, and it has much room left for improvement.

Regression analysis of the overall synergistic governance development level of haze emission

On the basis of clearly describing the spatial correlation network of haze emission in China and deeply analyzing the synergistic governance development of haze emission, this paper selects the annual value of the synergistic governance development level of inter-provincial haze emission as the explained variable, to respectively perform the ordinary least square (OLS) regression on the network density, network grade, and network efficiency of spatial correlation structure of haze emission in order to reveal the impact of the overall spatial correlation structure on the level of synergistic governance development. The regression results are presented in Table 3.
Table 3

The regression results of overall spatial correlation structure and synergistic governance development level

Explanatory variable

Network density

Network grade

Network efficiency

Constant term

0.560***

0.606***

0.647***

(0.000)

(0.000)

(0.000)

Regression coefficient

0.1283***

− 0.0449***

− 0.0839***

(0.002)

(0.002)

(0.004)

R 2

0.4283

0.4391

0.3950

*** Significant at 1% level

In accordance with Table 3, it can be seen that the overall spatial correlation structure has a significant effect on the level of synergistic governance development of haze emission in China and the increase of network density, the decline of network grade, and network efficiency will all enhance the synergistic development level.

The increase of network density indicates that the correlations among regions in the spatial network structure are strengthened, which helps to narrow the spatial difference of haze emission among regions. The decline of network grade means the grade system among regions in the spatial network structure is weakened and that the marginal regions will be able to play more roles in the synergistic governance of haze emission. The decline of network efficiency indicates that the number of inter-regional correlations and the number of inter-regional correlated channels in the spatial network structure increase, which helps to improve the synergistic governance development level of haze emission.

Conclusions and discussions

This paper comprehensively considers the important influence of “economic level” and “population scale” on “haze emission.” On the basis of the panel data of China from 1998 to 2016, the characteristics of spatial correlation network of haze emission and the level of synergistic development of haze governance in China are analyzed in detail. This paper explores the regions and cities that play a key role in the network, analyzes the network characteristics that affect the level of synergistic development, and designs effective ways to enhance the ability of China’s regional synergistic haze control. The main conclusions of this paper are as follows:

During the study period, the spatial correlation the spatial correlation of inter-provincial haze emission in China presented a typical “central–marginal” network structure. The network density increased continuously, the network grade and network efficiency continued to decline, the network connectedness remained stable, and the spatial correlation of haze emission among provinces in China was continuously strengthened. Most of the provinces in the network center are located in the Beijing–Tianjin–Hebei region and the Yangtze River Delta region of China. With the developed economy and mature environmental policies, they are in an advantageous position in the spatial correlation structure and play a leading role in the synergistic governance of haze emission in the whole country.

During the study period, the synergistic governance development level of inter-provincial haze emission in China showed a fluctuating upward trend, and the fluctuation amplitude was large. The synergistic governance development level of inter-provincial haze emission has increased from 0.589 in 1998 to 0.595 in 2016, with an overall increase of only 0.006. It suggests that the synergistic governance development level of inter-provincial haze emission in China is not high, the synergistic state is unstable, its overall growth rate is slow, and it has much room left for improvement.

More importantly, the overall spatial correlation structure has a significant effect on the level of synergistic governance development of haze emission in China and the increase of network density, the decline of network grade, and network efficiency will all enhance the synergistic development level.

Policy recommendations

It will be conducive to promoting spatial governance development of haze emission to enhance the spatial correlation of haze emission in various provinces in China, to narrow the spatial differences, and to give full play to the leading role of the central provinces in the spatial correlation network. Accordingly, the following policy suggestions can be drawn:

For the short-term planning, based on the “central–marginal” spatial correlation structure of the inter-provincial haze emission in China, full play should be given to the overall leading role of the central provinces in terms of haze governance, to establish a “leading–following” type of synergistic governance mechanism of haze emission. More specific, the Beijing–Tianjin–Hebei and the Yangtze River Delta regions are taken as the overall leaders of China’s haze governance, to be the pioneers in formulating and implementing various new environmental policies, and then adjust and improve these policies according to the implementation effect, and further spread them to other areas around the country, promoting the synergistic governance of haze emission.

For the medium-term planning, in consideration of the current instability of synergistic governance development level of haze emission in China, mechanisms should be explored to promote a stable synergistic pattern among local governments. At this point, the superior government should give full play to the role of macro-control to ensure the effective implementation of environmental policies, when the superior government can properly introduce the compensation mechanism or punishment mechanism to local governments. That is, local governments that benefit more from cooperative governance should compensate and pay part of the proceeds to other participants, or the superior government imposes administrative penalties on local governments that do not implement cooperative governance policies.

In the long-term planning, it is necessary to further reduce the grade of the spatial correlation network of haze emission, that is, to enhance the positions of the marginal provinces in the network. The main targets are the northeastern, central, and western Chinese provinces, where the superior government should introduce relevant policies to support the economic development, to alleviate the sharp contradiction between the economic development and environmental governance, and enhance the discourse power and status of the abovementioned provinces in the network structure. Finally, a fair, impartial, and equitable spatial correlation structure and synergistic governance of haze emission network can be established.

Limitations

The above conclusions are obtained on the basis of the combination of real data and theoretical models, but still have certain limitations. For example, when constructing the gravity model, only three indicators have been taken into consideration, that is, “haze emissions,” “economic level,” and “population scale.” In view of the complex actual situation, more relevant indicators should be incorporated into the model. In addition, after measuring the level of the synergistic governance development of haze, the deep factors affecting the level need to be further explored, which is the focus of the following-up study.

Footnotes

Notes

Acknowledgments

The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (71874189, 71774158, 71804182), China’s Post-doctoral Science Fund (2015M580484, 2016T90517), Jiangsu Social Science Fund (18JD013, 16JD008), High-Level Talents Project of “Six Talents Peaks” in Jiangsu Province (JY-077), and Qinlan Project of Jiangsu. We also would like to thank the anonymous referees for their helpful suggestions and corrections on the earlier draft of our paper, and upon which we have improved the content.

References

  1. Andersen ZJ, Raaschounielsen O, Ketzel M et al (2012) Diabetes incidence and long-term exposure to air pollution: a cohort study. Diab Care 35(1):92–98CrossRefGoogle Scholar
  2. Bertalanffy LV, 1969. General system theory: foundations, development, applications. G. BrazillerGoogle Scholar
  3. Borgatti SP, Mehra A, Brass DJ, Labianca G (2009) Network analysis in the social sciences. Science 323(5916):892–895CrossRefGoogle Scholar
  4. Carpenter A 2018. Oil pollution in the North Sea: the impact of governance measures on oil pollution over several decades. HydrobiologiaGoogle Scholar
  5. Cassi L, Morrison A, Terwal ALJ (2012) The evolution of trade and scientific collaboration networks in the global wine sector: a longitudinal study using network analysis. Econ Geogr 88(3):311–334CrossRefGoogle Scholar
  6. Chen YG, Liu JS (2002) Derivation and generalization of the urban gravitational model using fractal idea with an application to the spatial cross-correlation between Beijing and Tianjin. Geogr Res 21(6):742–752Google Scholar
  7. Chen H, Kwong JC, Copes R, Hystad P, van Donkelaar A, Tu K, Brook JR, Goldberg MS, Martin RV, Murray BJ, Wilton AS, Kopp A, Burnett RT (2017) Exposure to ambient air pollution and the incidence of dementia: a population-based cohort study. Environ Int 108:271–277CrossRefGoogle Scholar
  8. Ding G, Zhang B, Zhao P (2017) ESDA based spatial correlation features analysis of China urban regional haze pollution. Boletin Tecnico/Technical Bulletin 55(6):704–713Google Scholar
  9. Filonchyk M, Yan H, Li X (2018) Temporal and spatial variation of particulate matter and its correlation with other criteria of air pollutants in Lanzhou, China, in spring-summer periods. Atmospheric Pollution Research 9:1100–1110CrossRefGoogle Scholar
  10. Groenewold N, Guoping L, Chen A (2010) Regional output spillovers in China: estimates from a VAR model. Pap Reg Sci 86(1):101–122CrossRefGoogle Scholar
  11. Guo J (2014) Study on carbon emission and reduction measure of industrial sector based on network analysis method. Technol Econ 33(7):90–95Google Scholar
  12. Hart JE, Puett RC, Rexrode KM et al (2015) Effect modification of long-term air pollution exposures and the risk of incident cardiovascular disease in US women. J Am Heart Assoc 4(12):e002301CrossRefGoogle Scholar
  13. Head, K., 2003. Gravity for beginners. University of British ColumbiaGoogle Scholar
  14. Hou YH, Liu ZB, Yue ZG (2009) Social network analysis over the process of economic integration in the Yangtze River Delta. China Soft Sci 45(12):90–101Google Scholar
  15. Isard, W., 1956. Location and space-economy. M.I.T. Press 363Google Scholar
  16. Lee S (2017) Japanese measurement on fine particles (PM2.5) emission pollution and cooperation. Environ Resour Econ Rev 26(1):57–83Google Scholar
  17. Li L, Niu TY (2017) The spatial linkage network structure of regional innovation output—based on the social network analysis. Econ Geogr 37(09):19–25+61Google Scholar
  18. Li J, Chen S, Wan GH, Fu CM (2014a) Study on the spatial correlation and explanation of regional economic growth in China—based on analytic network process. Econ Res J 49(11):4–16Google Scholar
  19. Li HD, Wang S, Liu Y (2014b) Evaluation method and empirical research of regional synergetic development degree based on grey relational theory and distance collaborative model. Syst Eng-Theory Pract 34(7):1749–1755Google Scholar
  20. Li J, Fan C, Yuan Q (2017) Synergetic development degree evaluation of Beijing-Tianjin-Hebei region based on distance collaborative model. Sci Technol Manag Res 37(18):45–50Google Scholar
  21. Li BT, Guo HM, Wu HM (2018) Research on the influencing factors of ESCO dynamics in the development of existing building energy-saving reconstruction market based on SNA. Sci Technol Progress Policy 35(24):150–154Google Scholar
  22. Li H, Song Y, Zhang M (2018a) Study on the gravity center evolution of air pollution in Yangtze River Delta of China. Nat Hazards 90(3):1447–1459CrossRefGoogle Scholar
  23. Li M, Li C, Zhang M (2018b) Exploring the spatial spillover effects of industrialization and urbanization factors on pollutants emissions in China’s Huang-Huai-Hai region. J Clean Prod 195:154–162CrossRefGoogle Scholar
  24. Lin XQ, Wang D (2016) Spatio-temporal variations and socio-economic driving forces of air quality in Chinese cities. J Geogr Sci 26(11):1533–1549CrossRefGoogle Scholar
  25. Liu YB, Dai L (2013) Economic contact analysis with network structure of urban agglomeration around the Poyang Lake based on SNA. Resour EnvironYangtze Basin 22(3):263–271Google Scholar
  26. Liu H, Du G 2018. Research on spatial correlation of haze pollution in China. Statistical ResearchGoogle Scholar
  27. Liu HJ, Pei YF, 2017. An empirical test of the environmental Kuznets curve of China’s haze pollution. Stat Res (3): 45–54Google Scholar
  28. Liu HJ, Liu CM, Sun YN, et al., 2015. Spatial correlation network structure of energy consumption and its effect in China. China Ind Econ (5):83–95Google Scholar
  29. Newton I Sir, et al., 2007. Mathematical principles of natural philosophy. DawsonsGoogle Scholar
  30. Poyhonen PA (1963) Tentative model for the volume of trade between countries. Weltwirtschaftliches Arch 90:93–100Google Scholar
  31. Ravenstein EG (1885) The laws of migration. J Stat Soc Lond 48(2):167–235CrossRefGoogle Scholar
  32. Reilly WJ, 1929. Methods for the study of retail relationships//methods for the study of retail relationships. University of TexasGoogle Scholar
  33. Schiavo S (2010) International trade and financial integration: a weighted network analysis. Quant Finan 10(4):389–399CrossRefGoogle Scholar
  34. Shao S, Li X, Cao J, et al., 2016. China’s economic policy choices for governing smog pollution based on spatial spillover effects. Econ Res J (9):73–88Google Scholar
  35. Sun YN, Liu HJ, Liu CM, Cui R, 2016a. Research on spatial association of provinces carbon emissions and its effects in China. Shanghai J Econ (2):82–92Google Scholar
  36. Sun YN, Liu HJ, Liu CM, et al., 2016b. Research on spatial association of provinces carbon emissions and its effects in China. Shanghai J Econ (2):82–92Google Scholar
  37. Taaffe EJ (1962) The urban hierarchy: an air passenger definition. Econ Geogr 38(1):1–14CrossRefGoogle Scholar
  38. Tang FH, Tang H, Sun Q et al (2013) Analysis of the economic network structure of urban agglomerations in the middle Yangtze River. Acta Geograph Sin 68(10):1357–1366Google Scholar
  39. Tinbergen J 1962. An analysis of world trade flows in shaping the world economy. Edited by Jan Tinbergen New York NY: Twentieth Century FundGoogle Scholar
  40. Tong Z, Chen Y, Malkawi A, Liu Z, Freeman RB (2016) Energy saving potential of natural ventilation in China: the impact of ambient air pollution. Appl Energy 179:660–668CrossRefGoogle Scholar
  41. Van DA, Martin RV, Brauer M et al (2016) Global estimates of fine particulate matter using a combined geophysical-statistical method with information from satellites, models, and monitors. Environ Sci Technol 50(7):3762CrossRefGoogle Scholar
  42. Wang WQ 2014. The research on cooperative governance of air pollution:based on the case of Beijing area. J Public Manag (1):55–64Google Scholar
  43. Wang DZ, Zhuang RX, 1996. The preliminary probe into the quantitative analysis of regional economic links—a case study on economic links between Su-Xi-Chang and Shanghai. Sci Geogr Sin (01):51–57Google Scholar
  44. Wang X, Wu DT, Wang HQ, 2006. An attempt to calculate economic links between cities. Urban Stud (03):55–59Google Scholar
  45. Waters N 2014. Social network analysis. Springer Berlin HeidelbergGoogle Scholar
  46. Wu CY, Huang XJ, Chen BW et al (2017) Analysis of economic and spatial linkage and economic integration trend in Yangtze River economic belt from social network analysis perspective. Econ Geogr 37(07):71–78Google Scholar
  47. Xie Y, Zhao L, Jian X et al (2016) A cooperative reduction model for regional air pollution control in China that considers adverse health effects and pollutant reduction costs. Sci Total Environ 573:458–469CrossRefGoogle Scholar
  48. Yang WY (1989) Principle of location theory. Gansu People’s Publishing House, LanzhouGoogle Scholar
  49. Zhang M, Li H (2018) New evolutionary game model of the regional governance of haze pollution in China. Appl Math Model 63:577–590CrossRefGoogle Scholar
  50. Zhang DG, Lu YQ (2017) Study on the spatial correlation and explanation of carbon emission in China—based on social network analysis. Soft Sci 31(4):15–18Google Scholar
  51. Zhang Z, Hao Y, Lu ZN (2018) Does environmental pollution affect labor supply? An empirical analysis based on 112 cities in China. J Clean Prod 190(20):378–387CrossRefGoogle Scholar
  52. Zheng GH, 2010. Evaluation system on harmony development of regional innovation system. Chin J Syst Sci (3):76–79Google Scholar
  53. Zipf GK (1946) The P1P2/D hypothesis: on the intercity movement of persons. Am Sociol Rev 11(6):677–686CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2019

Authors and Affiliations

  1. 1.School of ManagementChina University of Mining and TechnologyXuzhouChina
  2. 2.Jiangsu Energy Economy and Management Research BaseChina University of Mining and TechnologyXuzhouChina

Personalised recommendations