EN中文

Journal of Geographical Sciences >

2019 , Vol. 29 >Issue 2: 231 - 252

DOI: https://doi.org/10.1007/s11442-019-1594-1

Research Articles

Spatial spillover effect and driving forces of carbon emission intensity at the city level in China

  • WANG Shaojian ,
  • HUANG Yongyuan ,
  • ZHOU Yuquan
Expand
  • Guangdong Provincial Key Laboratory of Urbanization and Geo-simulation, School of Geography and Planning, Sun Yat-sen University, Guangzhou 510275, China

Author: Wang Shaojian (1986-), Associate Professor, specialized in urban geography and regional development. E-mail:

Received date: 2018-04-07

  Accepted date: 2018-06-01

  Online published: 2019-02-25

Supported by

National Natural Science Foundation of China, No.41601151

Natural Science Foundation of Guangdong Province, No.2016A030310149

Pearl River S&T Nova Program of Guangzhou (201806010187)

Copyright

Journal of Geographical Sciences, All Rights Reserved
Fold

Abstract

In this study, we adopt kernel density estimation, spatial autocorrelation, spatial Markov chain, and panel quantile regression methods to analyze spatial spillover effects and driving factors of carbon emission intensity in 283 Chinese cities from 1992 to 2013. The following results were obtained. (1) Nuclear density estimation shows that the overall average carbon intensity of cities in China has decreased, with differences gradually narrowing. (2) The spatial autocorrelation Moran’s I index indicates significant spatial agglomeration of carbon emission intensity is gradually increasing; however, differences between regions have remained stable. (3) Spatial Markov chain analysis shows a Matthew effect in China’s urban carbon emission intensity. In addition, low-intensity and high-intensity cities characteristically maintain their initial state during the transition period. Furthermore, there is a clear “Spatial Spillover” effect in urban carbon emission intensity and there is heterogeneity in the spillover effect in different regional contexts; that is, if a city is near a city with low carbon emission intensity, the carbon emission intensity of the first city has a higher probability of upward transfer, and vice versa. (4) Panel quantile results indicate that in cities with low carbon emission intensity, economic growth, technological progress, and appropriate population density play an important role in reducing emissions. In addition, foreign investment intensity and traffic emissions are the main factors that increase carbon emission intensity. In cities with high carbon intensity, population density is an important emission reduction factor, and technological progress has no significant effect. In contrast, industrial emissions, extensive capital investment, and urban land expansion are the main factors driving the increase in carbon intensity.

Key words： Chinese cities; kernel density estimation; spatial autocorrelation; spatial spillover effect; spatial Markov chain; quantile regression panel model

Cite this article

WANG Shaojian , HUANG Yongyuan , ZHOU Yuquan . Spatial spillover effect and driving forces of carbon emission intensity at the city level in China[J]. Journal of Geographical Sciences, 2019 , 29(2) : 231 -252 . DOI: 10.1007/s11442-019-1594-1

1 Introduction

Although the proposal that greenhouse gases, especially carbon emissions, are the primary cause of climate change remains controversial, promoting an emission reduction plan to address the challenges posed by climate change is the consensus of mainstream scientific circles (Liu et al., 2008). In this context, China has achieved rapid and continuous economic growth after the reform and opening up, with the associated rapid advancement of urbanization and industrialization increasing demand for energy consumption (Wang et al., 2014a, 2014b). According to statistics released by the International Energy Agency (IEA, 2009), in 2007, China’s total carbon emissions exceeded the United States and ranked first in the world. Facing international public opinion pressure, the Chinese government adopted the emission reduction task as binding in the medium and long-term plans for national economic and social development in 2009 (Wang et al., 2017), promising to reduce China’s carbon emission intensity (CEI) by 40-45% compared with 2005 by 2020. However, the BP World Energy Statistical Yearbook (2017), released in 2017, shows that China’s total carbon emissions are still rising, from 20.9% in 2005 to 27.5% in 2014. For China, reducing the intensity of carbon emissions while maintaining sustained economic growth remains a big challenge (Wang et al., 2015; Wang et al., 2016a, 2016b). Therefore, research evaluating the spatio-temporal patterns and driving factors of CEI will be beneficial to further development of emission reduction measures.
CEI, as carbon emissions per unit of GDP, is a measure of the relationship between carbon emissions and economic development in a country or region. For developing countries, it is more practical to use CEI to quantify emission reduction targets compared to total carbon emissions or per capita carbon emissions (Jotzo and Pezzey, 2007). Many scholars have studied the issue of carbon emissions in the last decade with rich results, including measurement and decomposition, the characteristics of spatio-temporal patterns (including regional differences, spatial agglomeration, and spatial correlation), and driving factors of CEI (Zhao et al., 2018). The research scale varies from national, regional, and provincial to municipal units, although most have focused on the provincial level. In terms of research methods, scholars often use the Theil index, coefficient of variation, and spatial autocorrelation (the global Moran’s I index and local G coefficient) to reveal regional differences and spatial correlations of CEI. For example, Sun et al. (2012) explored differences in CEI among China’s provinces and regions (eastern, central, western, and northeastern) based on the Theil index. They found that China’s inter-provincial CEI varies and the differences have gradually increased, but intra-regional differences are declining annually, and inter-provincial differences are primarily derived from regional differences. Zhao et al. (2011) divided the country into eight economic regions and used the Theil index and global Moran’s I to explore the spatial evolution characteristics of regional CEI and also found a widening difference in regional CEI across the country caused by greater differences among regions, while intra-regional differences were small. In summary, there are significant differences in the CEI between provinces in China due to the variations in their geographical regions. However, few studies have focused on the city level. In addition, research regions are often divided according to habitual experience, such as central, eastern and western parts of China, the six economic zones, coastal areas, and interior land areas. In this regard, many studies have shown that barriers between regions have gradually decreased over time, and elements and products between adjacent cities and counties can move more freely (Li et al., 2004), which has gradually narrowed the scale of spatial spillover effects. As a result, analyses at the provincial scale alone do not address the true situation in China (Pan, 2012). In addition, most economic geographers believe that cities are key to regional development and the main source of carbon emissions and building low-carbon cities will be the most important component of carbon emissions reduction (Gu et al., 2009; Zhang et al., 2014), which highlights the importance of CEI research at the city level. Based on the limitations of previous research, this study evaluated spatial spillover effects and driving factors of CEI in 283 cities across China based on the city level. We further explain the spatial characteristic of CEI and explore its influencing factors.
Many scholars have introduced convergence theory and methods to conduct research on spatial and temporal characteristics of carbon emissions (Xu, 2010; Lin and Huang, 2011; Pettersson et al., 2014; Wu and Guo, 2016). Because the graphical features of the kernel density estimation method can reflect data evolution trends, Ma et al. (2015) and Zhao et al. (2018) used it to describe carbon emission performance and intensity over the years. At the same time, Zhao et al. (2017) introduced the Markov state transition probability matrix to reflect the internal dynamic evolution trend of CEI. However, the above method only explores changes in the region, ignoring the proximity and spatial characteristics of carbon emissions. Considering these two important characteristics, this study adds spatial lag conditions to the traditional Markov chain method and uses the spatial Markov chain transition probability matrix to dynamically reveal the spatial and temporal evolution process of CEI in cities. Therefore, we address the lack of regional background evaluation in the existing literature.
In terms of driving factors, scholars have primarily used exponential decomposition, structural decomposition, cointegration test, Granger causality test and regression empirical analysis based on panel data. Cheng et al. (2013) used spatial econometric models and found that energy structure, industrial structure, and urbanization rate play an important role in the evolution of CEI. Yan et al. (2016) used the Sharpe value decomposition method to decompose and analyze the CEI for 29 provinces in China and found that the contribution rate from economic development to the CEI difference is the largest and increases gradually over time. Zhou and Wang (2014) used the panel regression model to explore the driving factors of CEI and found that capital per capita, urbanization, marketization, and industrial structure are the main factors of the difference. In contrast, scholars have mainly discussed the influencing factors based on scale, structural, and technical effects. In addition, according to the International Energy Agency (IEA), in 2007, the global transportation sector emitted 66.23 tons of carbon dioxide, accounting for 23% of all energy activities. Therefore, transportation is one of the most important factors affecting carbon emissions. According to estimates by Cai et al. (2011) and others, China’s road transport accounted for 86.32% of transportation emissions in 2007, which is the main source of carbon emissions from transportation. Current research on the relationship between transportation and carbon emissions mainly includes the influencing factors of transportation carbon emissions (Zhang and Zeng, 2013), the impact of the transportation sector (Glaeser and Kahn, 2010), and transportation infrastructure (Xie et al., 2017) on carbon emissions. Based on existing research results (Li et al., 2018; Su et al., 2018), this study selected economic development level, population density, industrial structure, capital investment intensity, foreign capital intensity, land urbanization, and road density as driving factors, combining the quantile sorting technique in the Markov method and quantile regression method to explore the driving factors of urban CEI.
For these analyses, 283 cities in China were used as the research unit. First, the spatial autocorrelation method was used to test the spatial agglomeration of urban carbon emissions. Then, we used the spatial Markov chain to analyze the evolution of regional contexts and urban CEI and reveal the spatial spillover effects of urban CEI under different regional contexts. Finally, we used the panel quantile regression model to explore the driving factors affecting urban CEI under different regional backgrounds. The research design has three advantages. (1) The research scale is further reduced to the city level, which is in better agreement with the reality of spatial spillover effects. (2) Using the spatial Markov chain is effective at revealing the heterogeneity of spatial spillover effects for cities in different regional contexts. (3) Using quantile regression to explore the driving factors under different conditions of CEI level provides a basis for emission reduction strategies in cities with different CEI stages.

2 Data and methodologies

2.1 Research area and data sources

As the most concentrated area of human social and economic activities, cities have become the most concentrated carbon emissions regions (Gu et al., 2009). Therefore, many countries and regions regard cities as an important space carrier for local emission reduction measures (Cong et al., 2014) and promote building low-carbon cities. We use cities as the basic unit of research and define the urban scope using the administrative boundaries of each city in 2013. Due to statistics missing for some cities, 283 cities covering 34 provincial administrative units were studied, but Hong Kong, Macao, and Taiwan were excluded. The main variable studied was the CEI with a study period from 1992 to 2013.
In terms of data sources, due to the lack of detailed energy data for China’s cities, it is impossible to measure China’s urban carbon emissions comprehensively. To analyze the spatial spillover effects and driving factors of China’s urban CEI, urban carbon emissions data were compiled from published literature (Wang and Liu, 2017). City socio-economic data were obtained from China’s Urban Statistical Yearbook, China’s Regional Economic Statistical Yearbook, and the statistical yearbooks from various provinces, regions, and counties, Statistical Bulletin of National Economic and social development, and government work reports.

2.2 Research on spatial spillover effects based on spatial Markov chain

2.2.1 Markov chain
The traditional Markov chain is a discrete event stochastic process in mathematics, which is discrete in time and state, emphasizing that the historical state is not related to the future state. Because many geomorphic phenomena have no post-effects, the Markov chain is widely used in geography (Xu, 1996). Therefore, the Markov chain method is appropriate for evaluating CEI because urban CEI has no post-effects. Specifically, the Markov chain method discretizes the urban CEI at different time periods and divides a city’s CEI into k types according to quantiles. Then, the evolution process of urban CEI can be approximately regraded as a Markov process by calculating various types of probability distributions and its transition probabilities. In general, the urban carbon emission state type at time t is represented by a 1×k state probability vector Et=E1,t, E2,t,…, Ek,t. The state transition process for urban CEI over the entire study period can be represented by a k×k Markov transition probability, which is the matrix M. This study is based on the principle that each type of city has a similar CEI and divides the city’s CEI into four types according to quartiles (0.25/0.5/0.75), which are represented by k=1, 2, 3, and 4, respectively. A greater value of k is related to a greater intensity of carbon emissions. The state type from high intensity to low intensity is defined as an upward shift and from low intensity to high intensity is defined as a downward shift. mij represents the probability value of the region belonging to type i at time t, which transitions to type j at time t+1, and is estimated by the following equation:
${{m}_{ij}}=\frac{{{n}_{ij}}}{{{n}_{i}}}$ (1)
where nij represents the sum of the regions belonging to type i at time t that experiences transitions to type j at time t+1 during the entire study period, and ni is the sum of regions belonging to type i in all years of transition in the study period.
2.2.2 Spatial Markov chain
Regional connection and interaction processes create spatial spillover effects between regions, which reveal the spatial spillover effect between regions and are important in understanding regional development. The spatial Markov chain can better describe such regional spatial spillover patterns, and is essentially a product of the traditional Markov chain that introduces the concept of “space lag” (Gallo, 2004). From a geospatial perspective, regional phenomena are not isolated in geospatial space. Regional phenomena are always affected by the condition of geographically adjacent areas, that is, the state of regional context will have an impact on the state transition process of the region. The spatial Markov chain method expresses the regional context by introducing a “space lag”, thus addressing the deficiencies of the traditional Markov chain that ignored spatiality (Chen and Zhu, 2013). The spatial Markov chain transition probability matrix is based on the spatial lag type of region a at time t and decomposes the traditional Markov chain into k k×k conditional transition probability matrices. In the kth conditional matrix, mkij is the probability that a certain region a will shift from state type i at time t to the state type j at time t+1 under the condition that the spatial lag type is k. The spatial lag value for region a is the weighted average of the attribute values for a spatial neighboring area. The specific equation is as follows:
$\text{La}{{\text{g}}_{a}}=\mathop{\sum }^{}{{Y}_{b}}{{W}_{ab}}$ (2)
where the spatial weight matrix Wab represents the spatial relationship between region a and region b, and the subjacency principle is used to define the spatial relationship in this paper, due to missing statistics, for cities without neighboring cities, we define the closest city as the neighboring city. Yb represents the attribute value of region b, and Laga is the spatial lag value for region a, indicating the neighborhood state of region a.
By comparing the elements in the Markov transition probability matrix and spatial Markov transition probability matrix, the importance of the regional background to the regional transition can be judged.

2.3 Research on the driving factors based on quantile regression model

Most regression models have been developed using the classical least squares method, focusing on the influence that the independent variable x has on the conditional expectation E(y│x) of dependent variable y, which is essentially the mean regression and depicts the concentrated trend. However, the distribution of most data does not meet the classical assumption of the least squares method, and the conditional expectation E(y│x) poorly reflects the whole conditional distribution (Chen, 2010). To resolve this flaw, Koenker and Bassett (1978) proposed the “Quantile Regression”. The quantile regression characterizes the regression of the dependent variable using different independent variables quantiles, and the results can cover the influence of independent variables on the overall conditional distribution. In addition, this method uses the weighted average of the absolute values of the residuals as the objective function of minimization, so the estimation results are not affected by extreme values and are more stable than the least squares method. In the quantile regression, the τ quantile function Q(τ) of the explained variable y is defined as:
$Q\left( \tau \right)=inf\left\{ y:F\left( y \right)\ge \tau \right\}\text{ }\left( 0<\tau <1 \right)$ (3)
where τ represents the percentage of data below the regression line compared to the total data; therefore, the distribution of y is divided into two parts according to τ, the proportion less than quantile Q(τ) is τ, and the proportion greater than quantile Q(τ) is (1-τ).
For panel data, the general model is:
${{y}_{it}}={{x}_{it}}^{T}{{\beta }_{i}}+{{\alpha }_{i}}+{{u}_{it}}\text{ }\left( i=1,2,\text{ }\cdots \cdots ,K;\text{ }t=i=1,2,\text{ }\cdots \cdots ,\text{ }T \right)$ (4)
where i denotes different individuals, t denotes the time of sample observation, xit denotes the k×1-dimensional independent variable of the i-th individual in t period, u denotes a random error term vector, βi denotes an unknown coefficient of k×1-dimension, and αi represents the individual effect of the i-th individual. The conditional quantile function of panel quantile regression parameter estimation is:
${{Q}_{{{y}_{it}}}}\left( \tau \text{ }\!\!|\!\!\text{ }{{x}_{it}},{{\alpha }_{i}} \right)={{x}_{it}}^{T}\beta \left( {{\tau }_{q}} \right)+{{\alpha }_{i}}$ (5)
where τ$\in $(0, 1), and when τ takes different values, solving the weighted absolute residual minimization results in the parameter estimator at different quantile points. Parameter β is generally solved from the following equation:
$\hat{\beta}=argmin_{\alpha, \beta} \sum^{Q}_{q=1} \sum^{T}_{t=1} \sum^{N}_{i=1} w_{k} \rho_{\tau} (y_{it}-x^{T}_{it} \beta(\tau_{q})x_{it}-\alpha_{i})$ (6)
where ρτ(u) is a piecewise linear quantile loss function, and the specific expressions are:
${{\rho }_{\tau }}\left( u \right)=\left\{ \begin{matrix}u\cdot \left( \tau -1 \right),\text{ }u<0 \\u\cdot \tau ,\text{ }u\ge 0 \\\end{matrix} \right.$ (7)
In response to various problems in panel quantile estimation methods (Luo and Tian, 2010), Koenker proposed a penalty effect quantile regression method for fixed effects. The method appropriately adjusts the individual effect by adding the penalty term $P\left( \alpha \right)=\underset{i=1}{\overset{n}{\mathop \sum }}\,\left| {{\alpha }_{i}} \right|$| with the adjustment parameter λ to effectively reduce the variance caused by the estimation αi (Koenker, 2004). After adding the penalty term, the coefficient of the explanatory variable at the quantile can be obtained by solving the minimization problem using the following equation:
$\text{min}\underset{k=1}{\overset{q}{\mathop \sum }}\,\underset{t=1}{\overset{T}{\mathop \sum }}\,\underset{i=1}{\overset{N}{\mathop \sum }}\,{{w}_{k}}{{\rho }_{\tau k}}\left( {{y}_{it}}-{{x}_{it}}^{T}\beta \left( {{\tau }_{q}} \right){{x}_{it}}-{{\alpha }_{i}} \right)+\lambda \underset{i=1}{\overset{N}{\mathop \sum }}\,\left| {{\alpha }_{i}} \right|$ (8)
where wk is the weight coefficient used to control the degree of influence on the estimation coefficients of the q quantiles.
The factors affecting the intensity of carbon emissions have been explored in the published literature. The improved STIRPAT model based on the IPAT model proposes that the environmental pressure caused by human activities is mainly affected by population (P), affluence (A), and technological progress (T) and establishes a stochastic model between these three factors and the environmental impact (I) (York et al., 2003). The STIRPAT model is widely used in China to study various environmental impact indicators, such as carbon emissions, air pollution, and energy consumption. In recent years, many scholars have expanded the STIRPAT model. In addition to population, affluence, and technology, influencing factors include urbanization, economic growth, foreign trade, and industrial structure (Jiao and Chen, 2012). In addition, the hypothesis of the environmental Kuznets curve can be verified by adding a quadratic term or polynomial of wealth to the model. Based on the STIRPAT model and relevant CEI research results, this study analyzed the impact of human factors, such as affluence, population, industrial structure, foreign investment, urbanization, and road traffic on CEI (Table 1). Furthermore, we added the quadratic term (SA) for wealth to verify the environmental Kuznets curve hypothesis between wealth and environmental impact. It should be noted that, considering that government behavior values land urbanization, this study used land urbanization indicators to measure urbanization (Dong et al., 2018). CEI, urban affluence, population density, urbanization, and road density were placed logarithmically into the model.
Table 1 Main variables used in the applied model
Type Name Units Explanation
Explained
variable		 Carbon Emission Intensity (CEI) Tons / 10,000 yuan Total urban carbon emissions / GDP
Explanatory variables Affluence (A) yuan Per capita GDP
Population (P) (person/km2) Total population / city area
Industry Structure (IS) % Second industry added value / GDP
Investment Intensity (CI) % Total fixed assets investment / GDP
Foreign Direct Investment (FDI) 10,000 dollars / 10,000 yuan Actual use of foreign capital / GDP
Technology progress (T) 10,000 yuan / ton of standard coal Reciprocal of energy intensity (Total energy consumption / GDP)
Land urbanization (UB) % Road 500 m buffer area / city area
Road Density (RD) km/100 km2 Kilometer mileage / city area
In the model, for the null hypothesis “H0: all ui=0” in the fixed effect panel model and F(282,5935)=21.41 with a p value is 0.0000; therefore, the null hypothesis was rejected. Using the LSDV method for further investigation, most individual dummy variables were found significant (p=0.0000), so the null hypothesis that “all individual dummy variables are 0” could be rejected; that is, the model should not adopt the mixed regression model. Next, Prob > chi2 = 0.0000 was obtained using the Hausman test, which indicated that the model should use the fixed effect model. Therefore, the fixed-effect penalty quantile regression model was used to make estimates, and the R program package rqpd provided by Koenker was used to perform the calculation. The set estimation method was used to solve it, and the standard error was obtained through the bootstrap method (Koenker, 2004).

3 Empirical results and analysis

3.1 Time series and spatial correlation analysis of urban CEI

Before analyzing the spatial spillover effects and influencing factors, the time series and spatial correlation of urban CEI were analyzed. Figure 1 shows the spatial distribution of CEI in Chinese cities from 1992 to 2013. As shown, regions with higher CEI in 1992 were mainly concentrated in Heilongjiang, Jilin, Inner Mongolia, Ningxia, Shanxi, Shaanxi, Hebei, and Henan Provinces and the Pearl River Delta region. In contrast, by 2013, high CEI regions were mainly concentrated in Heilongjiang, Ningxia, and Shanxi Provinces. This distribution indicates that the overall CEI of Chinese cities has a downward trend, with particular prominence in the central, southern, and eastern regions.
Figure 1 CEI spatial patterns at the city level in China (1992-2013)
Figure 2 is a box-plot of urban CEI showing that during the 22-year study period, the average CEI of urban cities in China gradually decreased with differences between city shrinking and converging. To further understand the distribution of urban CEI, this study selected the four years, 1992, 2000, 2005, and 2013, for further kernel density estimates (Figure 3). The overall trend in urban CEI changed; the kernel density curve is a transition from “squat” to “high-thin” and the peak shifted to the left. This shows that both the overall intensity and overall gap in urban CEI decreased, indicating that the government’s carbon emission reduction measures have been effective. From the fluctuations at the end of the kernel density curve, between 2000 and 2013, numerical difference in cities with higher CEI gradually increased. This observation indicates that while the urban CEI decreased overall, a small part of the high intensity region has continued to miss the emission reduction target, and the gap with other cities increased.
Figure 2 Box-plot of CEI at the city level in China
Figure 3 Kernel density estimates of CEI at the city level in China
The box-plot and kernel density estimation methods only analyze the trend and distribution of urban CEI in the time dimension but cannot reflect the spatial characteristics of CEI. Therefore, the global Moran’s I index was adopted to portray the spatial characteristics of CEI of Chinese cities and test the spatial correlation. Table 2 shows the global Moran’s I index annual changes during the entire study period. All the years pass the test at the 1% significance level, indicating that urban CEI shows significant spatial clustering and positive spatial correlations in spatial distribution. Between 1992 and 2013, the overall Moran’s I index showed a weak upward trend with fluctuations that gradually decreased. This observation indicates that the spatial agglomeration of urban CEI gradually increased, and the energy utilization efficiency and technical level between adjacent cities had a certain spillover effect. Furthermore, the exchange and cooperation between cities and regions was generally stable while the level of spatial agglomeration decreased.
Table 2 Moran’s I for CEI at the city level in China (1992-2013)
Year Moran’s I Z value Year Moran’s I Z value
1992 0.359 8.717 ** 2003 0.434 11.018 **
1993 0.416 10.459** 2004 0.435 11.903 **
1994 0.336 8.807 ** 2005 0.498 11.564 **
1995 0.325 8.653 ** 2006 0.478 11.003 **
1996 0.292 7.565 ** 2007 0.464 12.150 **
1997 0.460 11.741 ** 2008 0.443 11.124 **
1998 0.316 8.143 ** 2009 0.489 10.880 **
1999 0.431 11.005 ** 2010 0.448 10.599 **
2000 0.411 10.504 ** 2011 0.435 9.534 **
2001 0.414 10.453 ** 2012 0.428 11.903 **
2002 0.417 10.516 ** 2013 0.380 11.564 **

Note: * indicates significant at the 5% level, and ** indicates significant at the 1% level

3.2 Analysis of spatial spillover effects of urban CEI

Urban CEI is divided into four types: low intensity, relatively low intensity, relatively high intensity, and high intensity, corresponding to k=1, 2, 3, and 4, respectively.
Figure 4 shows the spatial distribution of CEI in Chinese cities during the entire study period. The upward transfer of cities accounted for 57.95%, the downward transfer of cities accounted for 13.07%, and cities that remained stable accounted for 28.98% of the total number of cities. From the perspective of spatial distribution, the areas of upward transfer were mainly concentrated in northeastern and central China as well as provincial areas of Inner Mongolia, Shandong, and Guangdong. The downward transfer areas were concentrated in Guangxi, Yunnan, Shanxi, Ningxia, and southern Gansu Provinces, presenting a strong geographical agglomeration.
Figure 4 Spatial patterns of CEI changes in intensity types at the city level in China (1992-2013)
Table 3 shows the Markov transition probability matrix for the type of CEI in China. From the traditional Markov probability matrix, four important observations can be made.
Table 3 Markov matrix for CEI classes at the city level in China (1992-2013)
t/t+1 n 1 2 3 4
1 1439 0.896 5 0.098 0 0.005 6 0
2 1457 0.129 0 0.771 4 0.095 4 0.004 1
3 1512 0.001 3 0.171 3 0.779 8 0.047 6
4 1535 0.002 0 0.000 7 0.109 4 0.887 9
(1) The probability values on the diagonal of the probability matrix are larger than the probability values of the non-diagonal. The lowest value is 77.98%, and the highest value is 89.65%; that is, throughout the study period, the minimum probability of maintaining the original state is 77.98%, which indicates that the city’s CEI type has strong stability.
(2) On the diagonal, the stability of the relatively low intensity and relatively high intensity in the middle (77.14%, 77.98%) is significantly lower than the low and high intensity types at both ends (89.65%, 88.79%). From the probability values on both sides of the diagonal, the probability of the two types of upward transfer in the middle (12.9%, 17.13%) is greater than the probability of downward transfer in the middle (9.54%, 4.76%), which indicates that the relatively low intensity and relatively high intensity types show good momentum for upward transfer.
(3) There is a “Matthew effect” in the intensity of urban carbon emissions. In the type transfer process for consecutive years, the probability of the region maintaining low intensity is 89.65%, and the probability of a downward transfer is only 10.35%, which indicates that there is a convergence of the time dimension in the region with high carbon emission efficiency. The probability of maintaining an initial high-intensity type is 88.79%, and the probability of upward transfer is only 11.21%, indicating that the region may fall into the path of resource dependence and path locking, and it will be difficult to achieve energy efficiency.
(4) The non-diagonal and non-diagonal sides have a small probability value, and the maximum value is only 0.54%. Therefore, achieving technological progress and improving carbon emission efficiency is a continuous and gradual process. It is difficult to achieve rapid development in a short period of time, but at the same time, rapid development is not entirely impossible.
China’s urban CEI is not independent of geography. City CEI is often affected by regional location, with strong spatial agglomeration and spatial interaction effects (Lin and Huang, 2011; Wang et al., 2013). Concurrently, knowledge and technology spillovers have regional characteristics, and the spillover intensity and spatial distance attenuation have exponential function patterns (Wang et al., 2003). Figure 5 is a spatial distribution diagram of urban CEI type transfer after joining the urban neighborhood state. As shown, regions where the regional and neighborhood state types are an upward transition mainly concentrated in the northeastern and central China, as well as Shandong and Guangdong. The areas transferring downwards are mainly distributed at the junction of Shaanxi, Gansu, and Sichuan provinces, which shows clear geographical agglomeration.
Figure 5 Spatial patterns of CEI class transition for city units also showing neighborhood transitions in China (1992-2013)
The spatial Markov chain method can discern the influence of different neighborhood types on the probability of urban CEI class transition. Based on the traditional Markov transition probability matrix, the neighborhood type is added to obtain the spatial Markov transition probability matrix (Table 4). Assuming that the regional background (neighborhood type of the region) is not important for the transfer of the region, then the transfer probability matrix under different regional backgrounds should be equal and the corresponding elements in the traditional Markov transfer probability matrix should be equal. Comparing Table 3 and Table 4, we make four observations.
First, this assumption is not true. Based on different regional background conditions, the state transition of the region shows a large difference. Therefore, the regional background has a significant impact on the state transition of the region.
Table 4 Spatial Markov matrix for CEI classes at the city level in China (1992-2013)
Spatial Lag t/t+1 n 1 2 3 4
1 1 659 0.9408 0.0561 0.0030 0
2 357 0.1345 0.7703 0.0924 0.0028
3 257 0.0039 0.2257 0.7237 0.0467
4 176 0 0 0.1364 0.8636
2 1 478 0.8766 0.1192 0.0042 0
2 467 0.1370 0.7687 0.0878 0.0064
3 314 0 0.2070 0.7516 0.0414
4 204 0.0098 0.0049 0.1422 0.8431
3 1 244 0.8361 0.1516 0.0123 0
2 462 0.1364 0.7727 0.0887 0.0022
3 483 0.0021 0.1843 0.7723 0.0414
4 313 0.0032 0 0.1374 0.8594
4 1 58 0.8103 0.1724 0.0172 0
2 171 0.0760 0.7778 0.1404 0.0058
3 458 0 0.1026 0.8384 0.0590
4 842 0 0 0.0855 0.9145
Second, different neighborhood types have varying influences on regional state transitions. Generally speaking, if a region is adjacent to a region with low CEI, the probability of its CEI transferring upward will increase and transferring downward will decrease, and the neighbor will play a positive role in the regional state. If a region is adjacent to a region with high CEI, the probability of the CEI transferring upward will decrease and transferring downward will increase, and the neighbors will have a negative effect on the regional state. For example, for a region with high CEI, the probability of upward transfer is 11.21%, and in the low CEI background context, the probability of upward transfer increases to 13.64%. In the context of high intensity regions, the probability of an upward transfer is reduced to 8.55%. For regions with low CEI, the probability of downward transfer is 10.35%, when in the low CEI background context, the probability of downward transfer is reduced to 5.92%, while in the context of high intensity regions, the probability of its downward transfer increased to 18.97%.
Third, regional state types are affected differently by the regional background. For regions with lower CEI, when the neighborhood state types are 1, 2, and 3, the probabilities of upward and downward transfer are similar, about 13% and 9%. However, when there is an adjacent region with high CEI, the probability of upward transfer decreases to 7.6% and downward transfer increases to 14.62%. This observation indicates that areas with low CEI are more sensitive to neighbors with high CEI and are more easily negatively affected.
Fourth, from a dynamic perspective for the whole study period, 50.9% of the regions and the neighbors have the same state transition direction. The number of regions where the region and the neighbor type are simultaneously transferred upwards is 110, the number with downward transfer is 3, and the number of states that have not transferred is 31. Therefore, the regional and neighborhood CEI state transfer show synergy, with most having a coordinated upward transfer.
From this analysis, we conclude there is a spatial spillover effect on the CEI based on the regional background, and the type of spatial spillover has important significance. To test the statistical significance of this spatial spillover, a hypothesis test is required. The original assumption is that the types of CEI in the region are independent, regardless of the type of spatial lag. The model formula for the specific test is as follows:
${{S}_{b}}=-2log\left\{ \underset{l=1}{\overset{k}{\mathop \prod }}\,\underset{i=1}{\overset{k}{\mathop \prod }}\,\underset{j=1}{\overset{k}{\mathop \prod }}\,{{\left[ \frac{{{m}_{ij}}}{{{m}_{ij}}\left( l \right)} \right]}^{{{n}_{ij}}\left( l \right)}} \right\}$ (9)
where, k is the CEI type; mij is the traditional Markov transition probability, mij(l) and nij(l) represent the spatial Markov transition probability of the spatial lag type l and the corresponding number of cities; and Sb obeys the chi-square distribution with a degree of freedom of k(k‒1)2.
In the case where the degree of freedom is not adjusted, i.e., the element with a zero in the transition probability matrix throughout the study period is excluded, the degree of freedom is 4×(4‒1)2=36. From this formula, under the confidence level of α=0.005, ${{S}_{b}}=93.76>{{\chi }^{2}}(40)=66.77$. Therefore, we reject the assumption that China’s urban CEI class transfer is independent in space during the 1992-2013 period, and that there is a significant spatial correlation between the type of CEI and the type of state in the field.
Energy intensity is an important indicator of regional socioeconomic development, which has strong energy dependence; therefore, the spatial spillover effect pattern for energy intensity has important significance. The spatial Markov chain analyses show that the CEI in Chinese cities has significant spatial spillover effect and regional synergy. The spatial spillover effect shows two patterns: when in the high CEI region, the probability of rising local CEI increases, while in the low CEI region, the probability of a decline in local CEI increases. The results further illustrate the spatial correlation and spatial interaction of regional energy intensity. The regional synergy shows that the change in energy intensity in the region tends to be consistent, and a coordinated change of regional economic development leads to a coordinated change of energy consumption within the region. Concurrently, differing development patterns between regions show different energy needs and intensities.
Regional economic development is a complex system formed by the interaction of labor, capital, technology, and other elements in specific systems, resources, cultures, and other geographical environments (Zeng et al., 2015). Therefore, development has regional differences and intra-regional similarities. During marketization, the flow of factors between regions has caused various spillover effects, while payment transfers and technology diffusion have caused polarization and spillovers in China’s regional economic development (Long, 2003). The market potential theory of new economic geography points out that areas with high economic level and rapid development have great demand for products in surrounding areas, leading to a strong driving effect on them (Pan, 2012), and economic spillover effects make the energy demand between regions more spatially dependent. Moreover, transportation infrastructure is characterized by regional externality, with the network connecting many regions into one, which reduces the cost of factor flow (Zhang, 2012) and promotes the flow of inter-regional factors. More frequent factor flows require more energy to support but flows such as labor and capital strengthen the correlation in energy demand between regions. According to the first law of geography, the spatial correlation and dependence of local development are more manifested within the region. Under the conditions of the internal factor flow and improved interconnection infrastructure construction, the regional internal economic development pattern has gradually converged. In addition, technological spillover effects from exogenous and endogenous forces also affect regional energy intensity and characteristics (Li and Wang, 2008). The socio-economic activities of the region have complex spatial interaction processes, and patterns in spatial spillover effects of energy intensity are the result of the interaction of various factors, such as economic and institutional factors.

3.3 Analysis of driving forces of urban CEI

Although total carbon emissions in China have been increasing in the past two decades (1992-2013), the CEI has shown a downward trend, as the economic growth rate is generally higher than the growth rate of carbon emissions (Li et al., 2010; Zhang, 2010; Cheng et al., 2013). At present, China is in a transitional period of economic growth, dropping from 10.45% in 2010 to 6.9% in 2017. In this context, the nation’s emission reduction goals face a formidable challenge. The goal of analyzing the driving forces of urban CEI is to identify key factors that promote and reduce the intensity of carbon emissions from the urban level and identify effective measures for different types of cities.
To compare the mean regression coefficients of the traditional panel data model, we first estimated the common fixed effect panel model. In the panel quantile model estimation, five representative points, 10%, 25%, 50%, 75%, and 90%, were selected for estimation, and the total results are shown in Table 5. The panel quantile regression results show the variation in the elastic coefficients of all variables in the urban CEI distribution. We first focus on the impact of urban affluence on urban CEI. In both the mean regression and quantile regression, at a significance level of 1%, the logarithmic primary term for GDP per capita is positive and the secondary term is negative, indicating that there is an inverted “U-shaped” relationship between per capita GDP and CEI. Therefore, the intensity of carbon emissions in cities increases with the increase in per capita GDP. After the per capita GDP reaches a certain level, the CEI will decrease with increasing per capita GDP.
Table 5 Fixed effect and quantile regression estimates
Variables (1) (2) (3) (4) (5) (6)
FE q10 q25 q50 q75 q90
A 2.138*** 1.5515*** 1.6227*** 1.6329*** 1.4469*** 1.2669***
(8.85) (4.97) (5.58) (5.97) (5.16) (5.08)
SA -0.140*** -0.1018*** -0.1066*** -0.1102*** -0.1071*** -0.1023***
(-10.60) (-5.69) (-6.14) (-6.49) (-6.15) (-7.07)
P -0.210*** -0.0665** -0.0747*** -0.0792*** -0.0813*** -0.0771***
(-2.76) (-2.42) (-2.65) (-2.89) (-2.97) (-2.81)
IS 0.369*** 0.1758** 0.1963** 0.3458*** 0.4681*** 0.4350***
(3.21) (2.25) (2.16) (2.85) (4.88) (5.22)
CI 0.159 0.0982** 0.0954 0.1408 0.2881** 0.4066***
(1.65) (2.42) (1.36) (1.46) (2.53) (3.34)
FDI 5.427** 8.6024*** 7.7062*** 6.3488*** 4.9094*** 3.3846***
(1.99) (5.15) (4.85) (4.09) (3.84) (3.03)
T -0.0221*** -0.2610*** -0.2528*** -0.2087*** -0.0922 -0.0182
(-4.80) (-4.48) (-3.68) (-2.71) (-1.12) (-0.41)
UB 0.0426*** 0.0347*** 0.0377*** 0.0474*** 0.0672*** 0.0836***
(4.79) (4.77) (4.9) (5.4) (5.98) (6.77)
ROD 0.0435*** 0.0301*** 0.0248*** 0.0187*** 0.0092 0.00349
(6.04) (3.17) (3.06) (2.64) (1.42) (0.65)
Cons -6.114*** -4.4682*** -4.5841*** -4.3932*** -3.0907** -1.9161*
(-4.91) (-3.14) (-3.58) (-3.76) (-2.54) (-1.71)
N 6226 6226 6226 6226 6226 6226

Note: The t statistic is represented in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01

From the regression results, population density and technological progress can significantly reduce urban CEI. Energy is the basic guarantee for the living and production of urban residents, and demographic factors are closely related to urban energy demand. As a key factor in the demographic factor, population density mainly affects the CEI by changing the way people live and behave, with two impacts (Liu et al., 2017; Chai, 2013). First, some cities will benefit from the economies of scale and agglomeration effects of high urban population density, promoting the sharing of urban public service facilities, the formation of knowledge spillovers, labor pools, and specialized intermediates to enhance urban productivity and reduce energy consumption. However, too high an urban population density creates the problem of uneconomical agglomeration. Overcrowding leads to rising competition costs, traffic congestion, and excessive construction, operation and maintenance costs for excessive infrastructure demand (Cai and Sun, 2013; Chen and Yang, 2007). From the empirical results presented in this study, the population density has a significant negative effect on CEI and there is no clear uneconomical agglomeration phenomenon. The absolute value of the estimated coefficients for each quantile of population density is greater than the mean regression coefficient, indicating that the mean regression has a tendency to exaggerate the negative influence. As the quantile changes, there is a significant difference in the population density estimation coefficient. The absolute value of the quantile coefficient greater than 50% is significantly larger than the absolute value of the estimated coefficient for the first 50% quantile, and the estimated coefficient varies from -0.0813 for the 75% quantile to -0.0665 for the 10% quantile. The compact emission reduction effect from population is clear for cities with high CEI; for cities with lower CEI, appropriately increasing population density is not the preferred emission reduction measure.
In general, technological progress reduces energy intensity by increasing energy efficiency, that is, consuming less energy for the same GDP. However, technological progress has both positive and negative effects on energy consumption. While technological progress can improve energy efficiency, at the same time, the return effect means that technological progress stimulates economic activity and increase energy consumption, thereby offsetting energy savings due to efficiency gains. In addition, due to the threshold and time lag effect for technological progress (Li and Zhou, 2006; Li and Qu, 2012), in less developed areas, the emission reduction effect from technological progress may not be reflected. The empirical results presented in this study show that technological progress has a negative effect on CEI, and the negative effect of technological progress under high quantile (75% and 90%) is not significant, while other quantiles at the 1% confidence level show a significant negative effect. The absolute value of the estimated coefficient increases as the quantile decreases, and the negative effect of technological advancement reaches a maximum at the 10% quantile. Therefore, technological progress is the main emission reduction factor in areas with low CEI. The energy saved by technological progress is greater than the energy demand increased by economic growth. In areas with high CEI, the contribution of technological progress to emission reductions is not significant.
Factors including industrial structure, investment intensity, foreign capital intensity, urbanization, and road density have contributed to the increase in urban CEI. Prior work suggests that one of the driving forces for China’s carbon emission growth comes from industrial structure (Zhang, 2011). In a sample from this study, the number of cities in which the secondary industry added value accounted for more than 50% of the GDP, approximately 58% of the total cities in China in 2013, and China’s urban industrial structure is still dominated by the secondary, tertiary, and then primary industries. Furthermore, the internal structure of the secondary industry is also unreasonable, with high energy consumption, high emissions, low efficiency, and strong energy dependence (Zheng and Liu, 2011). Because estimated coefficients show significant positive values in both mean regression and quantile, the proportion of the secondary industry is an important factor leading to an increase in urban CEI in China. For different CEI condition quantiles, the estimation coefficient decreases with the decrease of the quantile point, which indicates that the secondary industry has a stronger promotion effect in areas with high CEI. Because most regions with high CEI are still in the stage of rapid industrialization, coal-led energy structure and extensive industrial development patterns have increased energy consumption. Regions with relatively low CEI are mostly economically developed regions. Environmental regulation, industrial transfer, and upgrading have generally improved the energy efficiency of the secondary industry (Xiao et al., 2014; Zhao and Qiu, 2014).
The intensity of investment also promotes the growth of carbon emissions. The estimated coefficient increases with the increase in the quantile, but it is not significant at the mean return and 25% and 50% quantile points. In areas with high CEI, investment-driven extensive economic development is one reason for the increase in carbon emissions (Guo, 2010).
The relationship between foreign investment and energy consumption in the literature has generally been discussed in conjunction with the “Pollution Haven Hypothesis” and “Pollution Halo Hypothesis”. The “Pollution Haven Hypothesis” is that FDI has transferred high-energy and high-pollution industries, which has led to an increase in carbon emissions to the host country. The “Pollution Halo Hypothesis” is that under domestic environmental regulation, FDI raises the technical level of the host country through knowledge and technology spillovers, which improve energy efficiency and reduce energy intensity. The empirical results from this study support the “Pollution Haven Hypothesis”. From the perspective of city level, the intensity of foreign investment has a significant positive effect on the intensity of carbon emissions. In terms of different conditional quantile levels, foreign investment intensity shows a stronger promotion effect on areas with low CEI. One possible explanation is that most regions with high CEI are those with relatively low economic and technical development. The emission reduction effect brought by FDI technology spillover offsets some of the increased energy consumption; low CEI, due to the high competitive pressure of FDI companies, has hindered the technology spillover of FDI (Li and Liu, 2011).
China’s urbanization is accompanied by urban land expansion and utilization type transformation, which has changed the urban carbon sink and carbon cycle process. Concurrently, urban land development, including infrastructure and building construction, will also bring subsequent energy needs (Zhao et al., 2009). Our empirical results show that land urbanization has a significant positive role in CEI, and the estimation coefficient increases with increases in the conditional quantile. In areas with high CEI, the positive effect of land urbanization reaches a maximum, showing that local governments relying on land finance and urbanization of extensive and contagious land development have brought enormous pressure on urban emission reduction (Tian, 2011).
A reasonable urban structure can reduce traffic congestion and improve traffic efficiency, thereby reducing emissions. However, our empirical results show that excluding the 75% and 90% conditional quantile levels, which are insignificant, road density plays a significant positive role in the CEI at a confidence level of 1%. These results are supported by existing literature (She et al., 2015). The estimated coefficient decreases with the decrease in quantile, indicating that the effect of road density is stronger in areas with lower CEI. One possible explanation is that transportation has become an important factor in increasing energy consumption in areas with low CEI (Pan et al., 2010). That is, the increase in road density in low CEI areas has improved urban accessibility, spurred demand for transportation, and increased energy consumption in the transportation sector (Park, 2014).
The heterogeneity of urban CEI driving forces is mainly affected by two factors. From the vertical perspective, the main driving forces of CEI vary at different stages of urban development. From the horizontal perspective, differences in urban development patterns will also cause differences in CEI driving forces. In the early and middle stages of urbanization, industrialization and urbanization are rapid with city expansion and development from the investment of a large amount of capital; the relatively primary industrial structure and extensive growth increase energy consumption. From the middle and late stages of urbanization, the speed of urbanization and industrialization slow down and the industrial structure becomes gradually optimized, which improves the city’s functions and attracts foreign investment. Under the combined effect of FDI technology spillovers and internal and external factors accumulated by local technology, energy efficiency gradually increases. However, due to the improvement of the traffic network and demand for economic development, the mobility between the elements increases, resulting in an increase in energy consumption in the transportation sector. In addition, the city’s development pattern also significantly affects the city’s CEI. For example, resource-based cities in Shaanxi, Shanxi, Inner Mongolia, and Heilongjiang rely on the development and utilization of resources to form an industrial cluster with high energy consumption and high emissions. Therefore, the industrial structure of these cities is still the main factor in carbon emissions. In more developed cities along the eastern coast, the rising proportion of service industries, industrial transfer, and upgrades have reduced the energy consumption of industrial structures.
This analysis shows that for cities with low CEI, economic growth and technological progress are key measures to reduce carbon emissions. In addition, appropriate population density and compact urban development patterns can also reduce emissions, while foreign investment intensity and traffic emissions are the main factors that increase CEI. For cities with high CEI, focusing on the development of compact cities, appropriately increasing population density, and exerting population size and agglomeration effects are important means of reducing emissions, while industrial emissions, extensive capital investment, and urban land spread are the main factors promoting CEI.

4 Discussion and conclusions

This study uses the kernel density estimation method and the spatial autocorrelation method to derive the temporal and spatial evolution pattern of CEI in 283 cities in China for the 1992-2013 period. The kernel density estimation results show that the overall CEI for cities decreased and the difference narrowed. In addition, the distribution of urban CEI does not follow a normal distribution, and the information contained at both ends of the distribution cannot be expressed by mean regression. Therefore, the quantile regression method is a better choice for exploring the driving forces of CEI. The spatial autocorrelation Moran’s I index indicates that there has been a significant spatial agglomeration of urban CEI that gradually increased, but differences between regions tend to be stable.
The spatial dynamic analysis process based on the Markov chain and spatial Markov chain shows that there is a Matthew effect in China’s urban CEI, and both low-intensity and high-intensity cities have shown a maintenance in the dynamic transfer process for neighboring years. Moreover, a “spatial spillover” effect in urban carbon emissions is clear. There are heterogeneity characteristics of spillover effects in different regional contexts, wherein low CEI neighbors can effectively increase the probability of CEI transferring upwards and vice versa. In addition, we find a trend in the regional convergence for urban CEI, and more than half the regions have the same dynamic direction of urban and regional backgrounds in adjacent years.
To further explore the mechanisms driving urban CEI, we used the quantile regression method to evaluate economic development, population density, industrial structure, capital investment intensity, foreign capital intensity, technological progress, land urbanization, and road density. We find that urban economic development level, population density, and technological progress are conducive to reducing urban CEI, while industrial structure, capital investment intensity, foreign investment intensity, land urbanization, and road density increase urban CEI. In cities with low CEI, economic growth and technological progress are key factors for reducing emissions. Appropriate population density and compact urban development patterns can also reduce emissions; while foreign investment intensity and traffic emissions are the main factors in increasing carbon emissions. In cities with high CEI, focusing on the development of compact cities, appropriately increasing population density, and exerting population size and agglomeration effects are important means of reducing emissions; industrial emissions, extensive capital investment, and urban land expansions are the main factors in increasing CEI. We propose that in areas with low CEI, foreign investment should be directed to low-energy, low-pollution, high-efficiency, high-tech industries, while at the same time exerting the knowledge spillover effect brought by foreign investment. Furthermore, focusing on road traffic energy consumption, improving transportation structure, and developing transportation energy-saving technologies can be used to achieve emission reduction goals. For regions with high CEI, optimizing industrial structure and improving capital utilization efficiency are key to further reducing CEI. However, local governments should diminish the disorderly expansion of cities due to excessive dependence on land finance.
More generally, this study finds that the spatial Markov chain method can effectively measure the spatial and temporal evolution process and patterns of CEI in cities and regions, and intuitively reveal the heterogeneity and regional characteristics of the “space spillover” effect of CEI. This method narrows the evolution scale to adjacent years, shows the continuing process of regional CEI evolution, and highlights regional transfer trends in the context of the neighborhood. In addition, the scale of research at the urban level can better explain the heterogeneity of urban development within provinces and the convergence of provincial junctions, providing a scientific basis for governments to formulate emission reduction strategies for different cities. Quantile regression analysis is an important mechanism for exploring driving forces that can identify more comprehensive explanations for CEI. This method emphasizes the heterogeneity of driving forces in the context of different CEI, thus avoiding the idealized mean regression model. The results of this research better reflect true conditions, and as such can be used to formulate targeted policy measures for various cities across the country.

The authors have declared that no competing interests exist.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Cai B F, Cao D, Liu L C et al., 2011. China transport CO2 emission study.Advances in Climate Change Research, 7(3): 197-203. (in Chinese)

[2]
Cai Y Y, Sun B D, 2013. Spatial dispersion of population and economic growth: Based on empirical analysis of megacities.Urban Insight, 27(5): 94-101. (in Chinese)Concentration of population generates economy of scale and economic growth,which most empirical studies have confirmed.But cities constantly expand,a series of urban issues become increasingly prominent,which influence urban development.Thus,it becomes the focus of study whether more agglomeration yields better outcome.This study explores whether the mega-cities has suffered agglomeration diseconomies from the perspective of population disperse.Through empirical research of 34 mega-cities,this study finds that disperse of population significantly promotes economic growth,which means agglomeration may produce adverse effects on economic growth.Besides,the effects of population disperse on economic growth showing inverted U-shaped trend,which means excessive disperse is not conducive to economic growth.

[3]
Chai Z X, 2013. Density effects, development level and China’s urban carbon dioxide emission.On Economic Problems, (3): 25-31. (in Chinese)

[4]
Chen L W, Yang K Z, 2007. Productivity, urban scale and economic density: An empirical study on the economic effect of urban agglomeration.Social Sciences in Guizhou, (2): 113-119. (in Chinese)

[5]
Chen P Y, Zhu X G, 2013. Regional convergence at county level in China.Scientia Geographica Sinica, 33(11): 1302-1308. (in Chinese)Regional inequality has always been a greatly significant issue to both academic enquiry and government policy. The recent literature on regional inequality implies that both scale and spatiality have been playing significant roles in the economic geographical processes. Despite the tremendous scholars interest in regional inequality in China since the late 1970s, the current research work ignores the spatial effects in studying China's regional inequality at the county level. Based on the per capita GDP data at the county level from 1998to 2009, this article analyzes the transition across different classes in the per capita GDP distribution, by a comprehensive application of Markov framework. For per capita GDP, this article distinguishes among 5 classes based on the average of all the 2 345 county units, computes their Markov chain matrix and spatial Markov chain matrix, and then analyzes the changing spatial patterns of the class transitions and their surrounding counterparts. The analysis concludes as follows. 1) The process of regional convergence in China has been characterized by"convergence clubs"since 1998. Furthermore, spatial polarization is discovered. The Markov chain matrix indicates that the richest and the poorest counties do not seem to change their relative position over time. The most affluent counties appear persistent he 92.9% probability of the richest remaining richest. And 94.1% probability of the poorest remaining poorest is the largest entry in the transition matrix. 2) The robustness of per capita GDP class transitions in China differentiates across the three regional belts. The counties in East China remain the most stable in changing their relative position. By contrast, their counterparts in Central China change their relative position more easily鈥33.38% of the 701 counties experiencing downward mobility locate in East China, 42.8% in Central China and 23.82% in West China. Among 443 counties experiencing upward mobility, 21.67% are in the East China, 41.31% in Central China and 37.88% in West China. 3)The maps of spatial Markov transitions show that the per capita GDP class transitions in China are greatly affected by their spatial neighbors. Lower-level units are negatively influenced by being surrounded by other lower-level units. Also, higher-level neighbors tend to prevent units from falling down in the per capita GDP distribution. 4) Per capita GDP class transitions in urban agglomerations, such as Changjiang River Delta and Zhujiang River Delta, remain stable, while their surrounding counterparts are active in changing their relative position. The class distribution map also shows"core-periphery"spatial structures in the urban agglomerations and their neighboring counties. More emphases should be paid to the poor county clusters in the eastern and central China, and to the spatial dynamics underlying the changing patterns of regional inequalities in China by the policy makers.

[6]
Chen Q, 2010. Advanced Econometrics and Stata Application. Beijing: Higher Education Press. (in Chinese)

[7]
Cheng Y Q, Zhang S Z, Ye X Y et al., 2013. Spatial econometric analysis of CEI and its driving factors from energy consumption in China.Acta Geographica Sinica, 68(10): 1418-1431. (in Chinese)The economic and social development has been facing with serious challenge brought by global climate change due to carbon emissions. As a responsible developing country, China pledged to reduce its carbon emission intensity by 40%- 45% below 2005levels by 2020. The realization of this target depends on not only the substantive transition of society, economy and industrial structure in national scale, but also the specific action and share of energy saving and emissions reduction in provincial scale. Based on the method provided by the IPCC, this paper examines the spatio-temporal dynamic patterns and domain factors of China's carbon emission intensity from energy consumption in 1997- 2010 using spatial autocorrelation analysis and spatial panel econometric model. The aim is to provide scientific basis for making different policies on energy conservation and carbon emission reduction in China. The results are shown as follows. Firstly, China's carbon emissions increased from 4.16 Gt to 11.29 Gt in 1997-2010, with an annual rate of 7.15%, which was much slower than that of annual growth rate of GDP(11.72%); therefore, China's carbon emission intensity tended to decline. Secondly, the changing curve of Moran's I indicated that China's carbon emission intensity from energy consumption has a continued strengthening tendency of spatial agglomeration at provincial scale. The provinces with higher and lower values appeared to be path-dependent or space-locked to some extent. Third, according to the analysis of spatial panel econometric model, it can be found that energy intensity, energy structure, industrial structure and urbanization rate were the domain factors that have impact on the spatio- temporal patterns of China's carbon emission intensity from energy consumption. Therefore, in order to realize the targets of energy conservation and emission reduction, we should improve the utilizing efficiency of energy, and optimize energy and industrial structure, and choose the low- carbon urbanization way and implement regional cooperation strategy of energy conservation and emissions reduction.

DOI

[8]
Cong J H, Liu X M, Zhao X R, 2014. Demarcation problems and the corresponding measurement methods of the urban carbon accounting.China Population, Resources and Environment, 24(4): 19-26. (in Chinese)This paper analyses and clarifies the ‘direct emissions and indirect emissions',‘organizational boundary emissions and geopolitical boundary emissions' and‘scope 1 emissions,scope 2 emissions and scope 3 emissions' as well as other 9 urban carbon accounting boundary defined methods in different angles. The indirect emissions methodologies in existing major city-scale frameworks are compared,and find the scope 2 emissions are all included in these city-scale standards while the scope 3 emissions varies different. The pros and cons of consumption-based and production-based carbon accounting are also analyzed respectively,Which find that the consumption-based accounting will be an important direction of future urban carbon accounting methods. This paper also introduces the measurement methods of the three scope emissions,and some questions should be noticed are put forward. To improve the city greenhouse gas emission inventory,this paper suggests that all kinds of organization for making city-scale standards should be united for a more clear classification and accounting process of scope 3 emissions. For China cities,carbon accounting boundary-setting needs to be flexibly choosen based on the objective. Strengthening the coordination between the cities on carbon emission reduction and giving a systematic review and revision for existing policies and measures are also suggested for cities' governments. The national competent authority,when decomposing emission reduction targets to the cities ' governments, first needs to make clear boundary and measurement methods. The emission reductions of aviation and power sectors should monitor their emissions.

DOI

[9]
Dong F, Yu B L, Hadachin T et al., 2018. Drivers of CEI change in China.Resources, Conservation and Recycling, 129: 187-201.

DOI

[10]
Gallo J L, 2004. Space-time analysis of GDP disparities among European regions: A Markov chain approach.International Regional Science Review, 27(2): 138-163.

DOI

[11]
Glaeser E L, Kahn M E, 2010. The greenness of cities: Carbon dioxide emissions and urban development. Journal of Urban Economics, 67(3): 404-418.Carbon dioxide emissions may create significant social harm because of global warming, yet American urban development tends to be in low density areas with very hot summers. In this paper, we attempt to quantify the carbon dioxide emissions associated with new construction in different locations across the country. We look at emissions from driving, public transit, home heating, and household electricity usage. We find that the lowest emissions areas are generally in California and that the highest emissions areas are in Texas and Oklahoma. There is a strong negative association between emissions and land use regulations. By restricting new development, the cleanest areas of the country would seem to be pushing new development towards places with higher emissions. Cities generally have significantly lower emissions than suburban areas, and the city-suburb gap is particularly large in older areas, like New York.

DOI

[12]
Gu C L, Tan Z B, Liu W et al., 2009. A study on climate change, carbon emissions and low-carbon city planning.Urban Planning Forum, (3): 38-45. (in Chinese)Climate change has increasingly influenced human life.Carbon emissions has become a major factor causing global climate warming.Related researches from China and overseas found that carbon emissions intertwined process of urbanization, low-carbon city became the first choice to mitigate global climate wanning.China is a socialist market economy country with rapid economic growth,accelerating urbanization and increasing carbon emissions,low-carbon city planning is a key technology for the development of low-carbon cities.This paper focuses on the relationship among climate change,carbon emissions and China's urbanization,and reviews new progress of low-carbon city planning researches from two aspects of the theoretical studies and the Action Plans.

DOI

[13]
Guo C X, 2010. An analysis of the increase of CO2 emission in China: Based on SDA technique.China Industrial Economics, (12): 47-56. (in Chinese)

DOI

[14]
International Energy Agency (IEA), 2010. World Energy Outlook 2009 Factsheet.

[15]
Jiao W X, Chen X P, 2012. Environmental impact analysis of Gansu province based on the stirpat model.Resources and Environment in the Yangtze Basin, 21(1): 105-110. (in Chinese)

[16]
Jotzo F, Pezzey J C V, 2007. Optimal intensity targets for greenhouse gas emissions trading under uncertainty.Environmental & Resource Economics, 38(2): 259-284.Uncertainty is an obstacle for commitments under cap and trade schemes for emission permits. We assess how well intensity targets, where each country permit allocation is indexed to its future realized GDP, can cope with uncertainties in international greenhouse emissions trading. We present some empirical foundations for intensity targets and derive a simple rule for the optimal degree of indexation to GDP. Using an 18-region simulation model of a cooperative, global cap-and-trade treaty in 2020 under multiple uncertainties and endogenous commitments, we show that optimal intensity targets could reduce the cost of uncertainty and achieve significant increases in global abatement. The optimal degree of indexation to GDP would vary greatly between countries, including super-indexation in some advanced countries, and partial indexation for most developing countries. Standard intensity targets (with one-to-one indexation) would also improve the overall outcome, but to a lesser degree and not in all individual cases. Although target indexation is no magic wand for a future global climate treaty, gains from reduced cost uncertainty and the potential for more stringent environmental commitments could justify the increased complexity and other potential downsides of intensity targets.

DOI

[17]
Koenker R, 2004. Quantile regression for longitudinal data.Journal Multivariate Analysis, 91(1): 74-89.The penalized least squares interpretation of the classical random effects estimator suggests a possible way forward for quantile regression models with a large number of “fixed effects”. The introduction of a large number of individual fixed effects can significantly inflate the variability of estimates of other covariate effects. Regularization, or shrinkage of these individual effects toward a common value can help to modify this inflation effect. A general approach to estimating quantile regression models for longitudinal data is proposed employing ? regularization methods. Sparse linear algebra and interior point methods for solving large linear programs are essential computational tools.

DOI

[18]
Koenker R, Bassett G, 1978. Regression quantiles. Econometrica, 46(1): 33-50.

DOI

[19]
Li G Z, Wang S, 2008. Regional factor decompositions in China’s energy intensity change: Base on LMDI technique.Journal of Finance and Economics, 34(8): 52-62. (in Chinese)In this paper,we use logarithmic mean Divisia index(LMDI) technique to decompose regional energy intensity changes for the period 1995-2005.We find that the dominant contributor to the decline of total energy intensity is the technological change within regions,which is shown as the accumulated energy intensity within region.

[20]
Li J X, Chen Y N, Li Zet al., 2018. Quantitative analysis of the impact factors of conventional energy carbon emissions in Kazakhstan based on LMDI decomposition and STIRPAT model.Journal of Geographical Sciences, 28(7): 1001-1019.Quantitative analysis of the impact factors in energy-related CO_2 emissions serves as an important guide for reducing carbon emissions and building an environmentally-friendly society. This paper aims to use LMDI method and a modified STIRPAT model to research the conventional energy-related CO_2 emissions in Kazakhstan after the collapse of the Soviet Union. The results show that the trajectory of CO_2 emissions displayed U-shaped curve from 1992 to 2013. Based on the extended Kaya identity and additive LMDI method, we decomposed total CO_2 emissions into four influencing factors. Of those, the economic active effect is the most influential factor driving CO_2 emissions, which produced 110.86 Mt CO_2 emissions, with a contribution rate of 43.92%. The second driving factor is the population effect, which led to 11.87 Mt CO_2 emissions with a contribution rate of 4.7%. On the contrary, the energy intensity effect is the most inhibiting factor, which caused –110.90 Mt CO_2 emissions with a contribution rate of –43.94%, followed by the energy carbon structure effect resulting in –18.76 Mt CO_2 emissions with a contribution rate of –7.43%. In order to provide an in-depth examination of the change response between energy-related CO_2 emissions and each impact factor, we construct a modified STIRPAT model based on ridge regression estimation. The results indicate that for every 1% increase in population size, economic activity, energy intensity and energy carbon structure, there is a subsequent increase in CO_2 emissions of 3.13%, 0.41%, 0.30% and 0.63%, respectively.

DOI

[21]
Li K J, Qu R X, 2012. Impact of technological change on carbon dioxide emission: An empirical analysis based on provincial dynamic panel data model.Journal of Beijing Normal University (Social Sciences), (5): 129-139. (in Chinese)Technological change is a key factor affecting carbon dioxide emission,and to analyze the impact of the technological change can provide basis to understand the variation and develop policies of emission reduction.This paper uses the data envelope analysis method to measure China's technological change,then investigates empirically the impact of technological change.The results show that the current and previous technological changes have negative impact on carbon dioxide emission,and the impact of current technological change is less than that of the previous technological change.Population and per capita GDP positively affect carbon dioxide emission,and the 1% change of population and per capita GDP may amount to 0.284% and 0.346% change.The effect of technological change represents some disparities.In the eastern and western regions,technological change can reduce carbon dioxide emission;however,technological change tends to increase it in the central region.Therefore,to achieve the emission reduction goal,China should attach importance to the role of technological change,promote low carbon technological change,strengthen the monitor and regulation of carbon dioxide emission,pay attention to the quality of growth,and restrict the approval and transfer of high pollution and energy-intensive industries.

[22]
Li L S, Zhou Y, 2006. Can technological progress improve energy efficiency? Empirical test based on China’s industrial sector.Management World, (10): 82-89. (in Chinese)

[23]
Li S T, Hou Y Z, Liu Y Z et al., 2004. The analysis on survey of local protection in China domestic market. Economic Research Journal, (11): 78-84. (in Chinese)

[24]
Li Y M, Zhang L, Cheng X L, 2010. A decomposition model and reduction approaches for carbon dioxide emissions in China.Resources Science, 32(2): 218-222. (in Chinese)CO2 emissions caused by more and more energy consumption which is largely dependent on coal have been increasing consistently in China. From 1953 to 2007, total CO2 emissions increased over quadruple. In general, the increasing process can be divided into four stages: a low growth stage during the period 1953 to 1980 with showing a fast but unsteady increasing rate and a small increasing magnitude; a steady growth stage during the period 1981 to 1996 with showing a steady increasing rate and a large increasing magnitude; a slowly decreasing stage during period 1997 to 2000 with indicating a slightly decreasing trend in CO2 emissions; a rapid growth stage during the period 2001 to 2007 with exhibiting a rapid increasing rate and a large increasing magnitude. In the present research, the authors made use of a factor decomposition model to quantify the effects of reductions in CO2 emissions resulting from three factors, i.e., growth of the economic total amount, evolution of industry structure, and variations in CO2 intensity during the period 1980 to 2007. Results showed that GDP and industry structure are the primary factors resulting in increases in CO2 emissions and decreasing intensity of CO2 emissions is the only factor causing effects on significant CO2 emissions. It was estimated that the contribution value of GDP was 1688 million during the period 1980-2000 and 1031 million during the period in 2001-2007; the contribution value of industry structure was 242 million during the period 1980-2000 and 78 million during the period 2001-2007; the contribution value of CO2 intensity was -1322 million during the period 1980-2000 and -54 million during the period 2001-2007. To this end, effective paths to substantially reduce CO2 emissions are adjustment of the industry structure and effectively decreasing CO2 intensity which depends largely on the change in energy consumption structure and improvements in energy utilization efficiency. As China is experiencing a middle stage of industrialization and urbanization, heavy industry inevitably becomes the dominant industry. Domestic demands for energy intensive products such as steel and cement tend not to decrease dramatically in the short-term. It was suggested that adjusting industry structure must mach with adjusting trade structure whose CO2 emissions are also considerable. Due to limitations in energy resources, capital and technology, China would not be able to fundamentally change energy consumption structure in the short-term. However, actively increasing oil and gas imports and developing renewable energy are considered long-term task. Meanwhile, China can develop coal clean technology to reduce environmental pollution and degradation. Although China energy intensity has been decreasing recently, the energy intensity of some energy intensive industries are still higher than developed countries levels. There is, however, large room to continuously decrease CO2 intensity by improving energy utilization efficiency for China.

DOI

[25]
Li Z H, Liu H H, 2011. FDI, technology progress and emission of CO2: Evidence from Chinese provincial data.Studies in Science of Science, 29(10): 1495-1503. (in Chinese)As the problem of Greenhouse effect is worsen,more and more people start to focus on the effect of FDI on CO2 emission.Based on the panel data of 30 Chinese provinces from 2000 to 2008,using the absolute indicators and comparative indicators which represent CO2 emission level,and classifying the whole county into eastern entral and western areas according to the economical and geographical factor,this paper investigates the influence of FDI on CO2 emission through technical channel.Results from the whole country show that,the influence of FDI on CO2 emission is lagged,FDI lagged one term has significant positive effects on CO2 emission in China.For diverse areas,FDI in eastern area has a significant positive effect on local CO2 emission;the influence of FDI in central area is unapparent;FDI has a negative influence on CO2 emission in western area.

[26]
Lin B Q, Huang X G, 2011. Evolution trend of China’s regional carbon emission under the gradient development model—Based on the perspective of spatial analysis.Journal of Financial Research, (12): 35-46. (in Chinese)

[27]
Liu Y H, Gao C C, Lu Y Y, 2017. The impact of urbanization on GHG emissions in China: The role of population density.Journal of Clean Production, 157: 299-309.Urbanization directly drives rural to urban population migration and indirectly causes west to east migration in China, two phenomena that may significantly impact China's greenhouse gas emissions given its huge population and vast difference between the west rural and east urban areas. We analyzed these two phenomena by turning emissions into a per capita term, and extending the impact from the traditional urbanization rate effect to include population density effect. Our results show that population density has actually been the dominant demographic player in changing per capita emissions for the past two decades in China, and its elasticity changed from positive in economically less-developed provinces to negative for the developed provinces. This study therefore provides a new perspective in the study of the relationship between urbanization and greenhouse gas emissions, and our results indicate that population density change should be taken into account to accurately assess the impact of urbanization.

DOI

[28]
Liu Y H, Ge Q S, He F N et al., 2008. Countermeasures against international pressure of reducing CO2 emissions and analysis on China’s potential of CO2 emission reduction.Acta Geographica Sinica, 63(7): 675-682. (in Chinese)After analyzing the situation of international negotiation on climate change and the pressure of reducing CO2 emissions China is faced with,the paper puts forward eight countermeasures to cope with the increasing pressure,integrating national development plans with advancement of science and technology at home and abroad. Of the countermeasures,five internal ones,such as optimizing energy structure,improving energy efficiency,encouraging a nation-wide energy-saving movement,augmenting terrestrial and oceanic sink for CO2,and recognizing production transfer,will reduce a total of 50.73 billion t of CO2 emissions after national development plans are realized. It will be a substantial contribution of China to the international mitigation effort and is also an effective answer to the international pressure. Two external measures,such as participating in scientific debate and bidding for a larger emitting share,will alleviate some international pressure and obtain more time and more space for China. The last one is related to science and technology innovation.

DOI

[29]
Long G Y, 2003. Understanding China’s recent growth experience: A spatial econometric perspective.Annals of Regional Science, 37(4): 613-628.

DOI

[30]
Luo Y X, Tian M Z, 2010. Quantile regression for panel data and its simulation study.Statistical Research, 27(10): 81-87. (in Chinese)The paper discusses fixed effects panel data model and gives three quantile regression estimates of the unknown parameters.Monte Carlo simulation study indicates that these quantile regression methods are effective in deal with the panel data model and do better than mean regression methods when the error distribution is non-normal.Finally,a real data is studied and some useful reference information for decision-making is obtained.

[31]
Ma D L, Chen Z C, Wang L, 2015. Spatial econometrics research on inter-provincial carbon emissions efficiency in China.China Population, Resources and Environment, 25(1): 67-77. (in Chinese)The paper measures the efficiency of inter-provincial carbon dioxide emissions from 1998 to 2011 by using minimum distance algorithm which is strongest frontier. The advantage of this approach is that the change of inputs or outputs is minimum after the efficiency reaching the production frontier. On this basis,we analyze the regional differences and spatial correlation of interprovincial carbon emissions efficiency. At last,we establish the spatial econometric model to make an empirical study on influence factors of carbon emissions efficiency by using panel data of 30 provinces from 1998 to 2011. The results show that: during the sample period,the efficiency of inter-provincial carbon dioxide emissions shows a large inter-provincial differences,and average carbon emission efficiency in the eastern coastal provinces is significantly higher than of it in the inland provinces. From the perspective of region,the trend of carbon emissions efficiency in the east is relatively stable,but the trend of carbon emissions efficiency in nationwide and midwest China shows a "U"curve. And the carbon emission efficiency in the east of China is significantly higher than of it in the midwest China. Spatial autocorrelation test Moran's I shows that the efficiency of inter-provincial carbon emissions has the characteristics of spatial correlation and a significant clustering tendency. While LISA space diagram indicates that the inter-provincial carbon emissions efficiency has not only the characteristics of spatial dependence,but also the performance of spatial heterogeneity.The economic scale,industry structure and energy consumption structure have a negative impact on carbon emission efficiency,while opening-up,enterprise ownership structure and government intervention have a positive impact on it. But industrial structure has no significant influence on carbon emission efficiency. Therefore,the focus of the work for future China to improve the carbon emissions efficiency is as below: realizing the economic growth mode from extensive to intensive,focusing on the adjustment of industrial structure and energy consumption structure,further enhancing the quality of opening-up and strengthening the government's efforts to reduce carbon emission.

[32]
Pan M, Lv B, Zhang C et al., 2010. Thinking of model system construction of building energy efficiency and green building.Urban Studies, (7): 6-11. (in Chinese)This paper analyses construction thinking of building energy efficiency and green building model system by means of model construction,model simulation and model application technology which are widely used in scientific research field.This paper discusses the significance and function of building energy efficiency and green building model system,then the composition,construction steps and methods of building energy efficiency and green building model system are mainly discussed.

[33]
Pan W Q, 2012. Regional linkage and the spatial spillover effects on regional economic growth in China.Economic Research Journal, (1): 54-65. (in Chinese)Based on the ESDA method,this paper explores the feature of regional linkage and the spatial correlation among China's regional per-capita GDP from 1998 to 2009.We find that there exists global spatial autocorrelation all over the country,and this kind of autocorrelation has been increasing since 1998.Meanwhile,the local spatial correlation is gradually being shown.During the last 10 years.With the help of the new-economy-geographical model,we try to explore the effect of the potential market on the regional economic growth in China.The empirical analysis shows that spatial spillover effect is important to the regional economic growth rate.1% increase of market potential raises GDP per capita growth by 0.47%.In terms of elasticity,the effect of market potential outperforms that of capital.However,the spatial spillover effect among regions will gradually vanish as the distance between regions increase.

[34]
Park J, 2014. The Effects of Compact City Form on Transportation Energy Consumption and Air Pollution. Beijing: Tsinghua University Press. (in Chinese)

[35]
Pettersson F, Maddison D, Acar S et al., 2014. Convergence of carbon dioxide emissions: A review of the literature.International Review of Environmental & Resource Economics, 7(2): 141-178.

[36]
She Q N, Jia W X, Pan C et al., 2015. Spatial and temporal variation characteristics of urban forms' impact on regional carbon emissions in the Yangtze River Delta.China Population, Resources and Environment, 25(11): 44-51. (in Chinese)Based on the regional land use information and carbon emission data from 1990 to 2010,this paper analyses the relationship between urban forms and regional carbon emissions in the Yangtze River Delta( YRD) where suffers from intensive human disturbance and rapid urbanization. Urban forms are characterized by landscape metrics including the size,shape,regularity,fragmentation and traffic coupling factor of urban patches. The results show that,1 the carbon emissions in YRD show significant spatial heterogeneities with high value in the middle,the southeast and northwest and low value in the southwest; 2 the carbon emissions increase rapidly from 1990 to 2010 with the average growth rate of 40. 55%( 1990- 1995),24. 47%( 1995- 2000),70. 71%( 2000- 2005) and51. 43%( 2005- 2010) respectively; 3 road density( RD),traffic coupling factor( CF) and the largest urban patch index( LPI)have the most significant impacts on regional carbon emissions which illustrated that urban forms affect the carbon emissions mainly through a wide variety of transportation; 4 the urban forms affect carbon emissions differently during the various stages of urbanization process. In the primary stage of urbanization,the growth of total urban areas and road density caused the increase of regional carbon emissions,while the increase of urban patch numbers,degree of polymerization and road buffer could reduce the regional carbon emissions. Compared with 1990 and 2010,the increasing mean euclidean nearest neighbor distance( ENN_ MN) in 2010 was negatively correlated with carbon emissions which could reduce the carbon emissions. Currently,it 's an efficient way from the perspective of urban forms by increasing the areas of road buffer,urban patches' nearest neighbor distance and reducing the areas of the largest urban patches to control the gross amounts of carbon emissions. The results could provide further objective understanding of urban forms and regional carbon emissions as an empirical case and significant reference for optimizing urban forms and controlling regional carbon emissions.

[37]
Su W S, Liu Y Y, Wang S J et al., 2018. Regional inequality, spatial spillover effects, and the factors influencing city-level energy-related carbon emissions in China.Journal of Geographical Sciences, 28(4): 495-513.Data show that carbon emissions are increasing due to human energy consumption associated with economic development. As a result, a great deal of attention has been focused on efforts to reduce this growth in carbon emissions as well as to formulate policies to address and mitigate climate change. Although the majority of previous studies have explored the driving forces underlying Chinese carbon emissions, few have been carried out at the city-level because of the limited availability of relevant energy consumption statistics. Here, we utilize spatial autocorrelation, Markov-chain transitional matrices, a dynamic panel model, and system generalized distance estimation(Sys-GMM) to empirically evaluate the key determinants of carbon emissions at the city-level based on Chinese remote sensing data collected between 1992 and 2013. We also use these data to discuss observed spatial spillover effects taking into account spatiotemporal lag and a range of different geographical and economic weighting matrices. The results of this study suggest that regional discrepancies in city-level carbon emissions have decreased over time, which are consistent with a marked spatial spillover effect, and a lub' agglomeration of high-emissions. The evolution of these patterns also shows obvious path dependence, while the results of panel data analysis reveal the presence of a significant U-shaped relationship between carbon emissions and per capita GDP. Data also show that per capita carbon emissions have increased in concert with economic growth in most cities, and that a high-proportion of secondary industry and extensive investment growth have also exerted significant positive effects on city-level carbon emissions across China. In contrast, rapid population agglomeration, improvements in technology, increasing trade openness, and the accessibility and density of roads have all played a role in inhibiting carbon emissions. Thus, in order to reduce emissions, the Chinese government should legislate to inhibit the effects of factors that promote the release of carbon while at the same time acting to encourage those that mitigate this process. On the basis of the analysis presented in this study, we argue that optimizing industrial structures, streamlining extensive investment, increasing the level of technology, and improving road accessibility are all effective approaches to increase energy savings and reduce carbon emissions across China.

DOI

[38]
Sun Y H, Zhong W Z, Qing D R, 2012. Analysis on differences of CEI of each province in China based on Theil index.Finance and Trade Research, 23(3): 1-7. (in Chinese)

[39]
Tian L, 2011. Urbanization of land in urbanization progress of China: Boon or bane?City Planning Review, 35(2): 11-12. (in Chinese)Since the establishment of Land Use Rights system in 1988,the urbanization of land has played a role of double-edged sword in the rapid growth of cities.On the one hand,it opened up local revenue and facilitated the city infrastructure investment.On the other hand,excessive reliance on land revenue contributed to the rocketing housing price and urban sprawl,and thus led to serious social,economic and environmental problems.This paper analyzes the merits and shortcomings of land urbanization,and concludes with policy implications and measures of transformation of land urbanization in the 12th Five-Year Plan.

[40]
Wang S J, Fang C L, Guan X L et al., 2014a. Urbanization, energy consumption, and CO2 emissions in China: A panel data analysis of China’s province.Applied Energy, 136: 738-749.Global warming resulting from rapid economic growth across the world has become a worldwide threat. The coordination of development of urbanisation, energy consumption, and carbon dioxide (CO2) emissions therefore forms an important issue; it has attracted considerable attention from both governments and researchers in recent years. This study investigated the relationship between urbanisation, energy consumption, and CO2 emissions over the period 1995 2011, using a panel data model, based on the data for 30 Chinese provinces. The potential to reduce CO2 emissions was also analysed. The results indicated that per capita CO2 emissions in China were characterised by conspicuous regional imbalances during the period studied; in fact, per capita CO2 emissions decrease gradually from the eastern coastal region to the central region, and then to the western region. Urbanisation, energy consumption, and CO2 emissions were found to present a long run bi-directional positive relationship, the significance of which was discovered to vary between provinces as a result of the scale of their respective economies. In addition, a bi-directional causal relationship was found to exist between urbanisation, energy consumption, and CO2 emissions: specifically, a bi-directional positive causal relationship exists between CO2 emissions and urbanisation, as well as between energy consumption and CO2 emissions, and a one way positive causal relationship exists from urbanisation to energy consumption. Scenario simulations further demonstrated that whilst China per capita and total CO2 emissions will increase continuously between 2012 and 2020 under all of the three scenarios developed in this study, the potential to achieve reductions is also high. A better understanding of the relationship between urbanisation, energy consumption, and CO2 emissions will help China to realise the low-carbon economic development.

DOI

[41]
Wang S J, Fang C L, Ma H T et al., 2014b. Spatial differences and multi-mechanism of carbon footprint based on GWR model in provincial China.Journal of Geographical Sciences, 24(4): 804-822.Global warming has been one of the major concerns behind the world's high-speed economic growth. How to implement the coordinated development of the carbon footprint and the economy will be the core issue of the world's economic and social development, as well as the heated debate of the research at home and abroad in recent years. Based on the energy consumption, integrated with the "Top-Down" life cycle approach and geographically weighted regression(GWR) model, this paper analyzed the spatial differences and multi-mechanism of carbon footprint in provincial China in 2010. Firstly, this study calculated the amount of carbon footprint of each province using "Top-Down" life cycle approach and found that there were significant differences of carbon footprint and per capita carbon footprint in provincial China. The provinces with higher carbon footprint, mainly located in northern China, have large economic scales; the provinces with higher per capita carbon footprint are mainly distributed in central cities such as Beijing, Shanghai and energy-rich regions and heavy chemical bases. Secondly, with the aid of GIS and spatial analysis model(GWR model), this paper had unfolded that the expansion of economic scale is the main driver of the rapid growth of carbon footprint. The growth of population and urbanization also acted as promoting factors for the increase of the carbon footprint. Energy structure had no considerable promoting effect for the increase of the carbon footprint. Improving energy efficiency is the most important factor to inhibit the growing carbon footprint. Thirdly, developing low-carbon economies and low-carbon industries, as well as advocating low-carbon city construction and improving carbon efficiency would be the primary approaches to inhibit the rapid growth of carbon footprint. Moderately controlling the economic scale and population size would also be required to alleviate carbon footprint. Meanwhile, environmental protection and construction of low-carbon cities would evoke extensive attention in the process of urbanization.

DOI

[42]
Wang S J, Fang C L, Wang Y, 2016a. Spatiotemporal variations of energy-related CO2 emissions in China and its influencing factors: An empirical analysis based on provincial panel data.Renewable & Sustainable Energy Reviews, 55: 505-515.This paper examines carbon dioxide (CO2) emissions from the perspective of energy consumption, detailing an empirical investigation into the spatiotemporal variations and impact factors of energy-related CO2 emissions in China. The study, which is based on a provincial panel data set for the period 1995 2011, used an extended STIRPAT model, which was in turn examined using System-Generalized Method of Moments (Sys-GMM) regression. Results indicate that while per capita CO2 emissions in China were characterized by conspicuous regional imbalances during the period studied, regional inequality and spatial autocorrelation (agglomeration) both decreased gradually between 1995 and 2011, and the pattern evolutions of emissions evidenced a clear path dependency effect. The urbanization level was found to be the most important driving impact factor of CO2 emissions, followed by economic level and industry proportion. Conversely, tertiary industry proportion constituted the main inhibiting factor among the negative influencing factors, which also included technology level, energy consumption structure, energy intensity, and tertiary industry proportion. Importantly, the study revealed that the CO2 Kuznets Curve (CKC), which describes the relation between CO2 emissions and economic growth, in fact took the form of N-shape in the medium- and long-term, rather than the classical inverted-U shape of the environmental Kuznets Curve (EKC). Specifically, an additional inflection appeared after the U-shape relationship between economic growth and CO2 emissions, indicating the emergence of a relink phase between the two variables. The findings of this study have important implications for policy makers and urban planners: alongside steps to improve the technology level, accelerate the development of tertiary industry, and boost recycling and renewable energies, the optimization of a country energy structure that can in fact reduce reliance on fossil energy resources and constitute an effective measure to reduce CO2 emissions.

DOI

[43]
Wang S J, Fang C L, Wang Y et al., 2015. Quantifying the relationship between urban development intensity and carbon dioxide emissions using a panel data analysis.Ecological Indicators, 49: 121-131.As a factor associated with urban management and planning, urban development intensity (UDI) could in fact form the basis for a new rationale in coordinating urban sustainable development and reducing CO2emissions. However, existing literature engaging in the task of quantifying the impacts of urban development intensity on CO2emissions is limited. Therefore, the goal of this study is to quantify the relationship between urban development intensity and CO2emissions for a panel made up of the five major cities in China (Beijing, Shanghai, Tianjin, Chongqing and Guangzhou) using time series data from 1995 to 2011. Firstly, this study calculated CO2emissions for the five selected cities and presented a comprehensive index system for the assessment of the level of urban development intensity based on six aspects (land-use intensity, economic intensity, population intensity, infrastructure intensity, public service intensity and eco-environmental intensity) using locally important socioeconomic variables. Panel data analysis was subsequently utilised in order to quantify the relationships between urban development intensity and CO2emissions. The empirical results of the study indicate that factors such as land-use intensity, economic intensity, population intensity, infrastructure intensity and public service intensity exert a positive influence on CO2emissions. Further, the estimated coefficients suggest that land-use intensity is the most important factor in relation to CO2emissions. Conversely, eco-environmental intensity was identified as having a major inhibitory effect on CO2emission levels. The findings of this study hold important implications for both academics and practitioners, indicating that, on the path towards developing low-carbon cities in China, the effects of urban development intensity must be taken into consideration.

DOI

[44]
Wang S J, Li Q Y, Fang C L et al., 2016b. The relationship between economic growth, energy consumption, and CO2 emissions: Empirical evidence from China.Science of the Total Environment, 542: 360-371.61The nexus between economic growth, energy use and CO2emissions for China examined.61Cointegration tests suggest presence of long-run relationship among the variables.61Generalized impulse response due to the external shocks to the system examined.61Bi-directional causality between economic growth and energy consumption.61Unidirectional causality from energy consumption to CO2emissions.

DOI PMID

[45]
Wang S J, Liu X P, 2017. China’s city-level energy-related CO2 emissions: Spatiotemporal patterns and driving forces.Applied Energy, 200: 204-214.

DOI

[46]
Wang S J, Liu X P, Zhou C S et al., 2017. Examining the impacts of socioeconomic factors, urban form, and transportation networks on CO2 emissions in China’s megacities.Applied Energy, 185: 189-200.In addition to socioeconomic factors, urban planning and transportation organization are beginning to play an increasingly important role in the reduction of CO2emissions. However, little attention has been paid to the ways in which this emerging role can be framed. Therefore, this study aims to examine the combined impacts of socioeconomic and spatial planning factors on CO2emissions in cities that have experienced rapid urbanization, using an econometric model and a comprehensive panel dataset incorporating socioeconomic, urban form, and transportation factors for four Chinese megacities eijing, Tianjin, Shanghai and Guangzhou, n the period 1990 2010. Making use of remote sensing land-use data, the digitization of transportation maps, and a set of socioeconomic data, we developed an extended STRIPAT model in order to empirically estimate the impacts of the selected variables on CO2emission levels in these cities. The results indicate that the socioeconomic factors of economic growth, urbanization, and industrialization will lead to increased CO2emissions, while the service level and technology level can contribute to the reduction of CO2emissions. The results also suggest that the expansion of urban land use and increases in urban population density should be controlled through urban planning measures in order to reduce CO2emissions. In addition, pursuing compact urban development patterns would also help to reduce CO2emissions. Transportation factors including urban road density and the traffic coupling factor were both found to have exerted significant negative effects on CO2emission levels, indicating that increases in the coupling degree between urban spatial structure and traffic organization can also contribute to reducing such emissions. Our results cast a new light on the importance of practices of urban planning and spatial optimization measures in achieving CO2emission reductions. The findings obtained in this study are seen as providing important decision support in building low-carbon cities in China.

DOI

[47]
Wang Y, Cheng X, Yin P H et al., 2013. Research on regional characteristics of China’s carbon emission performance based on entropy method and cluster analysis.Journal of Natural Resources, 28(7): 1106-1116. (in Chinese)The main influencing factors of carbon emission performance include economic structure, energy consumption structure, regional division of work and technical level of high-carbon industry. Although these indicators are in a certain degree of influence on COemissions per unit of GDP, it is difficult to measure the contribution of each indicator. In this paper, we selected the four main indicators mentioned, according to the entropy method to measure the weight of each indicator, applying multi-indicators cluster analysis methods under the index system, and dividing the whole country into 7 typical areas. The analysis result shows: the characteristic of the regional division of work, which has the largest weight of 0.4567, is the main factor resulting in regional differences in China provincial carbon emissions performance, then the energy structure and the technical characteristics of high carbon industry take the second place. The indicator of economic structure, with a weight of 0.0971, has a limited contribution to explain the regional differences at the provincial level of carbon emissions. In northern regions, the high share of high-carbon production market and low technical level results in low carbon emission performance. Especially in North China, the high proportion of coal in the energy consumption has a negative effect on carbon emission performance; in southeast coastal areas, due to its advanced technology, the negative effect from high share of high-carbon production has been offset to some degree; although the share of high-carbon production in central and western China is much lower than in the eastern regions at present, yet the technical level is low. The carbon emission performance decreased with a backward technical level. There are differences among the historical development of provinces; natural resource endowment and regional division of work, complying with the macroeconomic development of objective laws and economic location theory, in addition to a small number of developed areas, the other regions in China are difficult to change their economic and energy structure in a short period of time. The low-carbon policies should be focused on improving the technical level of high-carbon industries.

[48]
Wang Z, Ma C F, Wang Y et al., 2003. A geographical investigation into knowledge spillovers between regions.Acta Geographica Sinica, 58(5): 773-780. (in Chinese)

[49]
Wu J X, Guo Z Y, 2016. Research on the convergence of carbon dioxide emissions in China: A continuous dynamic distribution approach.Statistical Research, 33(1): 54-60. (in Chinese)This paper uses the continuous dynamic distribution approach to study the dynamic evolution and long-term trend of carbon emission intensity and per capita carbon emissions,basing on the panel data of 286 cities and prefecturelevel cities from 2002 to 2011 in China. It shows that both carbon dioxide emission intensity and per capita dioxide emissions are unimodal distributions during the study period,while from the trend,they will converge to a few "clubs"which are not determined by regional characters. To avoid the possible polarization of club convergence,the government should pay more attention to the cities with carbon dioxide emission intensity value of 4. 05,4. 7 times to yearly average or per capita carbon dioxide emissions of 3. 9,5. 0 times to yearly average.

[50]
Xiao Y F, Wan Z J, Liu H G, 2014. An empirical study of carbon emission tranfer and carbon leakage in regional industrial transfer in China: Analysis based on inter-regional input-output model in 2002 and 2007.Journal of Finance and Economics, 40(2): 75-84. (in Chinese)In global trade carbon pollution effects of international industrial transfer such as implicit carbon emission,carbon transfer and carbon leakage have drawn extensive attention from domestic and foreign scholars.However,at present China experiences a key stage of industrial transfer from coastal areas to central and western areas,and related carbon pollution research resulting from inter-regional industrial transfer has not attracted corresponding attention.Based on input-output approach and the basic data of inter-regional input-output tables in 2002and 2007in China,this paper quantitatively evaluates the size,direction and industry concerning export-oriented and consumption-oriented industrial transfer in eight regions.Furthermore, based on sub-region and sub-industry carbon emission coefficients in 2007,it studies the carbon emission transfer and carbon leakage effects caused by regional industrial transfer and further discusses the effect of industrial transfer on regional carbon emission.It arrives at the following conclusions:by industrial transfer in eastern coastal areas,northwest,northeast and other regions become the heavy disaster areas of carbon emission and carbon leakage,and in Beijing,Tianjin and northern coastal areas, industrial transfer has carbon emission reduction effect,so industrial transfer has differentiated carbon emission effects in regions,namely greater effects in northwest,and northeast areas and smaller effects in eastern coastal areas,Beijing and Tianjin.Therefore,the formulation of stricter environmental regulation and differentiated regional carbon emission reduction policy is imminent.

[51]
Xie R, Fang J Y, Liu C J, 2017. The effects of transportation infrastructure on urban carbon emissions.Applied Energy, 196: 199-207.Against the background of global warming, China faces the dual pressures of economic structural transformation and carbon emission reduction. While promoting economic development, the development and construction of transportation infrastructure has contributed to urban carbon emissions. Using an improved STIRPAT model, we examine panel data for 283 cities between 2003 and 2013 to explore the effects of transportation infrastructure on urban carbon emissions. The results show that transportation infrastructure increases urban carbon emissions and intensity. In addition, while the population scale effect of transportation infrastructure is conducive to decreasing carbon emissions, the economic growth and technological innovation effects of transportation infrastructure increase carbon emissions. Results also demonstrate that in large and medium-scale cities, construction of transportation infrastructure increases carbon emissions. In small cities, this relationship is not significant. Robustness tests support all findings. These results indicate that the effective development of carbon-abatement policies requires an examination of the effects of transportation infrastructure.

DOI

[52]
Xu G Y, 2010. The convergence in carbon dioxide emissions: Theoretical hypotheses and empirical research in China.The Journal of Quantitative & Technical Economics, (9): 31-42. (in Chinese)In this paper,the convergence model of carbon emissions is constructed with the method of convergence of economic growth analysis.On this basis,the convergence and divergence of per capita carbon emissions is studied empirically,using 1995~2007 provincial panel data in China.The results show that:①China's per capita carbon emissions are not β absolute convergence,but β conditional convergence and the three Club Convergence of the eastern,central and western regions;②The decline in the proportion of secondary industry and coal consumption contributes to the convergence of per capita carbon emissions,per capita income levels,clean technology and government's macro-environmental regulation behavior have the different impact in the different regions.Accordingly,this paper presents relevant policy implications.

DOI

[53]
Xu J H, 1996. Mathematical Methods in Contemporary Geography. Beijing: Higher Education Press. (in Chinese)

[54]
Yan Y M, Wang Z, Wu L Y et al., Analysis of the determinants of CEI on regional differences. Acta Scientiae Circumstantiae, 36(9): 3436-3444. (in Chinese)

[55]
York R, Rosa E A, Dietz T, 2003. STIRPAT, IPAT and ImPACT: Analytic tools for unpacking the driving forces of environmental impacts.Ecological Economics, 46(3): 351-365.Despite the scientific consensus that humans have dramatically altered the global environment, we have a limited knowledge of the specific forces driving those impacts. One key limitation to a precise understanding of anthropogenic impacts is the absence of a set of refined analytic tools. Here we assess the analytic utility of the well-known IPAT identity, the newly developed ImPACT identity, and their stochastic cousin, the STIRPAT model. We discuss the relationship between these three formulations, their similar conceptual underpinnings and their divergent uses. We then refine the STIRPAT model by developing the concept of ecological elasticity (EE). To illustrate the application of STIRPAT and EE, we compute the ecological elasticities of population, affluence and other factors for cross-national emissions of carbon dioxide (CO 2) from fossil fuel combustion and for the energy footprint, a composite measure comprising impacts from fossil fuel combustion, fuel wood, hydropower and nuclear power. Our findings suggest that population has a proportional effect (unitary elasticity) on CO 2 emissions and the energy footprint. Affluence monotonically increases both CO 2 emissions and the energy footprint. However, for the energy footprint the relationship between affluence and impact changes from inelastic to elastic as affluence increases, while for CO 2 emissions the relationship changes from elastic to inelastic. Climate appears to affect both measures of impact, with tropical nations having considerably lower impact than non-tropical nations, controlling for other factors. Finally, indicators of modernization (urbanization and industrialization) are associated with high impacts. We conclude that the STIRPAT model, augmented with measures of ecological elasticity, allows for a more precise specification of the sensitivity of environmental impacts to the forces driving them. Such specifications not only inform the basic science of environmental change, but also point to the factors that may be most responsive to policy.

DOI

[56]
Zeng G, Shang Y M, Si Y F, 2015. The convergent evolution of China’s regional economic development models.Geographical Research, 34(11): 2005-2020. (in Chinese)As regional economies develop, regional economic development models are also evolving dynamically. Due to the differences of society, economy, resources and environment,economic development models evolve differently. In this paper, we establish a theoretical framework to classify economic development models according to three dimensions: elements(resource-driven or human-capital-driven), systems(government or market), and relationship(endogenous or exogenous), based on thorough review on the existing theoretical discussion.And we identify eight kinds of economic development model, such as "Resource-MarketEndogenous" and "Resource-Government-Exogenous" models(the paper explains the meanings of the models). Then taking 16 typical economic models in China as an example, we analyze the evolutional process. We find that:(1) regional economic development models are evolving dynamically under the influence of path dependence and path creation.(2) During the evolution process of China's regional economic development models, the intangible elements,market and exogenous forces are becoming increasingly important. There has also been an obvious evolution from models such as "Resource-Government-Endogenous", "ResourceGovernment-Exogenous", and "Human capital-Government-Exogenous" towards "Human capital-Market-Exogenous" on the whole.(3) With mutual learning and extensive communication, regional economic development models in China have seen an evolution from differentiation to convergence. It should be noted that this convergence is mainly characterized by goal convergence, although the process is different. Elemental structure, system environment and endo-exogenous relationships have always been important aspects of regional economic development model research, however, most of which focused on one of the three aspects. This paper puts forward a comparatively thorough theoretical framework to examine all Chinese development models. It is an important supplement and development to the existing research.

[57]
Zhang G Y, 2010. Economic development pattern change impact on China’s CEI.Economic Research Journal, (4): 120-133. (in Chinese)

[58]
Zhang L F, 2011. Relations among the industry structure, energy structure and carbon emissions.Journal of Arid Land Resources and Environment, 25(5): 1-7. (in Chinese)The green house gas emission have made a notable impact on the economy,environment and mankind existence by the humanity activitiy.With the fasten economic development,the population increase,the increasingly energy growth,the carbon emissions are growing.There has a huge pressure to control the green house gases emissions.In order to lessen the green house gases emissions and to realise the market in 2020 and the economic sustainable development,the relationship among the industry structure,energy produce structure,energy consume structure,industrial energy consumption structure and the carbon emissions were analyzed,and the concrete measures were put forward.

DOI

[59]
Zhang L J, Liu G L, Qin Y C, 2014. Multi-scale integrated assessment of urban energy use and CO2 emissions. Journal of Geographical Sciences, 24(4): 651-668.Accurate and detailed accounting of energy-induced carbon dioxide (CO2) emissions is crucial to the evaluation of pressures on natural resources and the environment, as well as to the assignment of...

DOI

[60]
Zhang T X, Zeng A Z, 2013. Spatial econometrics analysis on China transport carbon emissions.Urban Development Studies, 20(10): 14-20. (in Chinese)Based on 1995 ～ 2010 statistical data of China's 28 provinces,this paper adopts the method of spatial econometrics to analysis the inherent mechanism of transport carbon emissions behind the spatio-temporal pattern and the evolution trend in China City Size rapid expansion process,using the convergence theory of economic growth to analysis the distribution characteristics and the convergence of the transport carbon emissions of China' s 28 provinces. The results are as follow: ①There is a spatial agglomeration tendency on China's transport carbon emissions,the same average spontaneous level of transport carbon emissions in the provinces,but there is a big difference in levels of spontaneous carbon emissions,high transport carbon emissions for the developed coastal areas,and the western regions is relatively low. ②From 1995 to 2010,China's transport carbon emissions there is no uniform σ convergence,but there are 16 stages( years) σ convergence and absolute β convergence,and the β convergence speed is about 8. 3%. ③The longterm equilibrium elasticity of the per capita GDP,urban built-up area and per car energy consumption on transport carbon emission are about 0. 93,0. 34,0. 65. Based on the research,China's transport carbon emission reduction policy orientation is put forward.

[61]
Zhang X L, 2012. Has transport infrastructure promoted regional economic growth? With an analysis of the spatial spillover effects of transport infrastructure.Social Sciences in China, (3): 60-77. (in Chinese)在综合考虑多维要素对中国区域经济增长的协同作用的基础上, 构建交通基础 设施对区域经济增长的空间溢出模型, 利用1993-2009年的中国省级面板数据和空间 计量经济学的研究方法, 实证分析得出以下主要结论。 (1) 中国交通基础设施对区 域经济增长的产出弹性值合计约0.05-0.07, 表明其对中国区域经济增长具有重要的作 用。 (2) 中国交通基础设施对区域经济增长的空间溢出效应非常显著, 若不考虑空间 溢出效应, 会高估交通基础设施对区域经济增长的作用。 (3) 外地交通基础设施对本 地经济增长表现为以正的空间溢出效应为主, 但是也有空间负溢出的证据。 (4) 在影 响区域经济增长的多维要素中, 劳动力与其他公共部门资本存量对中国区域经济弹性 的贡献仍然较大, 新经济增长因素与新经济地理因素的作用也不容忽视。 关键词: 交通基础设施 空间溢出 区域经济增长 空间计量模型 Taking full account of the synergistic effects of multidimensional factors on regional economic growth in China, this paper constructs a model of the spatial spillover effects of transport infrastructure on regional economic growth. Using provincial panel data from 1993 to 2009 and employing spatial econometric techniques, our empirical analysis comes to the following conclusions. (1) The total output elasticity of transport infrastructure for regional economic growth varies between 0.05 and 0.07, indicating its important role in such growth. (2) Transport infrastructure has very clear spatial spillover effects on regional economic growth; its role in regional economic growth will be overestimated if these are neglected. (3) For a specific region, transport infrastructure in other regions has mainly positive spillover effects on economic growth, but there is also evidence of negative spillover effects. (4) Among multidimensional factors contributing to regional economic growth, labor plus capital stock from other parts of the public sector make the greatest contribution to regional economic growth in China, followed by the new economic growth factors and new economic geography.

DOI

[62]
Zhao G M, Chen L Z, Sun L C et al., 2017. Markov steady state prediction of CEI in China, based on the perspective of spatial differentiation.Science and Technology Management Research, 37(22): 228-233. (in Chinese)

[63]
Zhao Q Z, Yan Q Y, Zhao H R, 2018. Research on spatial characteristics and influencing factors of provincial carbon emissions in China.Journal of Beijing Institute of Technology (Social Sciences Edition), 20(1): 9-16. (in Chinese)Kernel density distribution and Moran's Index methods were utilized to indicate the dynamic evolution trend and spatial cluster characteristics of carbon emissions among 30 provinces in China during 2000-2015. Spatial Durbin Model was constructed to explore the key influencing factors. The results are as follows:(1)Carbon emission density keeps a decreasing trend in this period and the low transformation trend has been accelerated since the New Normal Stage;(2)Spatial cluster characteristics of carbon emissions density in 30 provinces are mainly divided into "High-High"and "Low-Low"types. Moreover, this spatial spillover effects show a growing trend;(3)The economic scale and industrial structure of a province have a significant positive effect upon its carbon emission density while patent output scale has a significant negative effect. FDI scale and energy consumption structure of its neighborhood exert spatial spillover effects on its emission density significantly. On the one hand, to accelerate the pace of industrial structure adjustment, to optimize industrial spatial layouts and to develop green technology are the main ways in the future to stimulate regional low carbon transformation in China. Meanwhile, ecological town construction and continuously improving FDI quality are the potential factors to drive carbon emissions down. Spatial spillover effects among provinces in carbon emission reductions shouldn't be neglected.

[64]
Zhao R Q, Huang X J, Xu H et al., 2009. Progress in the research of carbon cycle and management of urban system.Journal of Natural Resources, (10): 1847-1859. (in Chinese)Land use change and fossil fuel combustion caused by urbanization are the major reasons for global climate change and greenhouse effect.Understanding the process,way,direction and mechanism of carbon cycle of regional urban system is helpful to better forecast the future concentration of atmospheric greenhouse gases and to put forward corresponding carbon management measures.Urban system is an integrated multi-factor and multi-level social-economic system,and its carbon cycle process has some characteristics that are quite different from natural ecosystems,such as complexity,uncertainty and spatial heterogeneity.The carbon cycle of urban system is a very complex process that includes natural and human process,horizontal and vertical process,economic and social process etc.,which are essentially different from that of natural ecosystems.Firstly,based on the carbon cycle research of urban system in China and overseas,this paper summarized and analyzed the characteristics,spatial extent and carbon cycle process of urban system.Secondly,the paper discussed the main research aspects in carbon cycle of urban system at present such as carbon emission from urban energy use,carbon cycle of urban vegetation and soil,the influence of urban expansion on carbon emission,city metabolism and carbon process,modeling carbon cycle of urban system etc.During the discussion of the above aspects,this paper mainly illustrated the key points and train of thought of each research field,and analyzed the effects of human factors and economic activities on the spatial heterogeneity of urban carbon cycle.Thirdly,it talked about the theoretical framework,scientific problems,main strategies and aims of carbon management according to URCM(Urban and Regional Carbon Management) research project.Finally,it put forward the future direction and suggestion in this field,such as strengthening the researches on economic and social mechanism of carbon cycle process of urban system,discovering the carbon cycle characteristics of different land use types and the effects of land use change on carbon emission,building carbon cycle model of urban system and adopting feasible carbon management practices to attain the aim of decarbonized city.

DOI

[65]
Zhao R Y, Qiu Z Z, 2014. Review on the relationship between industrial structure and carbon emission.Economic Review, (10): 110-113. (in Chinese)

[66]
Zhao Y T, Huang X J, Zhong T Y et al., 2011. Spatial pattern evolution of CEI from energy consumption in China.Environmental Science, 32(11): 3145-3152. (in Chinese)

[67]
Zheng C D, Liu S, 2011. Industrial structure and carbon emission: An empirical analysis based on China provincial panel data.Research on Development, (2): 26-33. (in Chinese)

[68]
Zhou J Q, Wang T S, 2014. Convergence of regional economic growth and CEI difference and its mechanism: An empirical analysis based on Chinese provincial panel data.Social Science Research, (5): 66-73. (in Chinese)

Options
Outlines

/

〈 〉