Analysis of Regional Difference and Correlation between Highway Traffic Development and Economic Development in China

Abstract

This study used three-tier architecture to analyze the relationship between the highway traffic development difference and the economic development difference in China from 1997 to 2015. Their time distribution and trends of national and regional differences were quantified via both coefficient of variation and Theil index. Furthermore, the highway traffic connectivity was analyzed for each province. The obtained results showed that the difference of highway traffic development was significantly related to the difference of economic development, and poor connectivity of highway traffic was also strongly related to economic disparity. The investment difference of China’s highway infrastructure could be divided into an unstable stage (1997–2010), and a stable stage (2011–2015). The highway traffic development difference could also be divided into two stages (1997–2006 and 2007–2015). Following the construction of central and western regions in around 2006, this difference decreased significantly; however, large differences within region remained. Provinces in the central and western regions have low connectivity and several remote main cities are not connected via high-grade highways. China’s national economic difference has been constantly decreasing; however, the three regions follow different trends. Furthermore, with the development of the central and western regions, between-region economic differences decreased rapidly. To improve the level of regional equalization for highway development, the government should reduce differences between developed and undeveloped provinces for each region and connect remote cities, thus ultimately reducing differences in economic development and promoting a balanced regional development.

Transportation construction has been identified as one of the major drivers of the economic growth of China during the past decade ( 1 ). However, China has extensively pursued a biased development of its transportation infrastructure, concentrating the main portion of public investment on the eastern regions, which directly caused the observed diversity of regional economic growth in China ( 2 , 3 ). During the “12th Five-Year Plan”, which considered the difference between poor and wealthy regions, the government has formulated a different transport development strategy, which generally reflects a fairer approach. In “The plan of poverty alleviation for traffic construction in the contiguous poverty-stricken areas (2011–2020),” the government has formulated the main direction toward promoting the equalization of transportation services, which laid a solid foundation for the achievement of common prosperity. This study investigated the disequilibrium between highway infrastructure development and economic development at three levels (national, regional, and provincial).

The enthusiasm of the economist to discuss the critical function of infrastructure investment on economic development was inspired during the late 1980s ( 4 ). Since then, considerable research emerged that examined the relationship between highway investment and the resulting economic growth. A significant share of the relevant literature used a production function model to estimate the impact of public capital stock on the resulting output of multiple economic sectors ( 5 , 6 ). Output elasticities have been reported to vary widely, which is likely due to differences in utilized spatial levels of analysis, definitions of capital stock, estimation techniques, or underlying models ( 7 ). Various studies have been conducted that specifically examined spillover effects ( 5 , 8 ) by dividing the economic impact of the transportation infrastructure into either direct and indirect effects ( 9 ) or internal and external effects ( 10 ). Berechman et al. ( 11 ) used three models and retrieved output elasticities of 0.37, 0.34, and 0.01 for state, county, and municipality levels, respectively, with respect to the obtained public highway capital. However, all these models and studies failed to take reverse causality into account, consequently overestimating the infrastructure variable coefficients in the production function ( 12 ). Furthermore, by adopting vector auto-regression method and structural equation modeling, many studies tested whether road investments would lead to productivity growth or whether productivity growth entails increased road construction ( 13 , 14 ). In other words, long-term research has investigated the causal relationship between transportation and regional economic growth.

In general, economically underdeveloped areas have been associated with poor infrastructure ( 15 ). In Europe, differences in the provision of transport infrastructure display a similar geographical pattern with the gross domestic product (GDP) per capita ( 16 ). Due to the inequality of the regional economic development in China, the exploration of the relationship between this uneven transportation development and the disparity of regional economic development has become a hot topic. It is therefore necessary to conduct an in-depth analysis of the disparity trend of regional economic growth, and to discuss how to effectively utilize the means of infrastructure investment and policies to promote the coordinated development of the regional economy.

Numerous studies examined whether the transportation development of China had differential effects on different parts of the country: Wang verified the uneven distribution of regional railway construction as the main reason for the strong differences in the economic growth rate between eastern, central, and western regions ( 2 ). Yu et al. calculated the output elasticities of the infrastructures of the eastern, central, and western regions and found that the central region has the highest elasticity, and further suggested that infrastructure investment in the central region should receive more attention by the central government ( 17 ). Zhang reported that both the transportation infrastructure and the economic growth showed strong spatial concentration characteristics and followed a decreasing trend from eastern to western regions ( 18 ). Wang et al. utilized exploratory spatial data statistics to analyze the spatial distribution difference of the highway network of northwest China ( 19 ). Cao et al. adopted the Natural Breaks function of the ArcMap software to divide transportation development levels of 183 cities from 1991 to 2003 and found that the level of urban transportation development was distinctly correlated to the per capita industrial output at the 0.01 level ( 20 ). Demurger studied the differences in the development of the inter-provincial transport network, using the density of the road network, and analyzed whether underdeveloped infrastructure networks likely led to a growing regional disequilibrium in China ( 21 ). Fan and Chan-Kang analyzed the relationships between road investment, economic growth, and poverty reduction in China, concluding that areas for road investment should be chosen to balance economic growth and to reduce poverty. The authors further indicated that investing in the coastal areas would boost economic growth, while investing inland (especially in the southwestern region) will reduce poverty ( 3 ).

Apparently, transport improvement strongly impacts regional development. A previous study reported that the implementation of a national trunk highway system has been divided into two phases. During the first phase (1990–2000), the general trend aggravates the division of Chinese economic regions, although regional disparity would be mitigated to some extent during the second phase (2000–2020) ( 22 ). The Chinese government committed to a substantial highway investment to benefit a balanced economic growth and to improve equalization, consequently defining poor and remote areas as target areas. However, the trend of regional differentiation in China, and the missing knowledge on how to effectively use highway construction to alleviate this trend, continues to provide major problems for policy makers to achieve a coordinated development of the regional economy. It is therefore important to investigate the relationship between highway traffic development and economic development at different levels and for different regions, as this is of great significance for the formulation of reasonable traffic policies and for the eventual promotion of a sustained economic development.

Regional differences can be described via the coefficient of variation (CV) as well as via Theil’s inequality index. As an index of diversity, the CV has been widely applied to the fields of economics and sociology to investigate and compare social inequality across nations, cities, and other social aggregates ( 23 , 24 ). The Theil index is additively decomposable and satisfies several desirable properties as a measure of regional inequality ( 25 ). This decomposition of the global index in partial factorial contributions is interesting in political terms as well as due to its analytical implications; it furthermore clarifies the global role or contribution of each part ( 26 ). We chose the CV and the first order decomposition of the Theil index to further study the differences between China’s highway investment and the economic development during the past years (1997–2015). The spatial correlation dimension was used to calculate the development of highway networks and the connectivity among main cities, thus testing the correlation between the connectivity of traffic networks and the economic difference within provinces. Via an analysis of the national, regional, and provincial level over time, this study exposed differences between transportation development and economic development, which will help policy makers to make more informed decisions to achieve the equalization of the highway traffic development.

This paper is organized as follows: First, the available data and the data source is described. Next, a framework for analysis and the utilized method is presented. The trends of highway investment growth as well as the regional differences within China are reviewed, and China’s highway construction differences and economic differences are analyzed. Finally, the relationship between connectivity and economic differences of the main cities for each province are presented. The paper concludes with a summary of the key findings and a discussion of necessary policy amendments.

Data and Data Source

This study chose 31 provincial administrative units of China (identical to the provinces used in this paper) as study area. According to the level of economic and technological development and its geographical location, the study area can be divided into three economic zones (identical to the economic regions used in this paper). Figure 1 shows the per capita GDP distribution and adjacency relationships of provinces and regions in the research area of China.

Figure 1.

The data used for this study are the investments for the completion of highway construction at the provincial level between 1997 and 2016, and at the national level between 1980 and 2016. These data were obtained from the Compilation of National Statistical Data on Traffic and Transportation; the per capita GDP, the permanent resident population, and each grade highway mileage at the provincial level between 1997 and 2015 were supplied from the National Bureau of Statistics, Statistical Yearbook of China; the per capita GDP of prefecture-level cities was obtained from the China City Statistical Yearbooks of 2006, 2012, and 2016; the spatial distance between the main cities in each province was calculated via ArcGIS; and the traffic distance between main cities was obtained from China’s Highway and Urban and Rural Road Atlas published in 2012. The atlas provides trip distances on the basis of data measured by 1200 drivers, who prefer to travel higher-grade highways. It should be noted that according to decreasing quality level, highways in China are divided into freeway, Grade I, Grade II, Grade III, and Grade IV; therefore, freeways should be selected first, followed by Grade I highways and so on. The sources of the same category data were identical, and all of these are official publications. Table 1 presents statistical summaries of regions for per capita GDP, permanent resident population, and highway mileage.

Table 1.

Summary Statistics of Regions of China

	1997	2005	2006	2011	2015
Provincial level per capita GDP (yuan)
Average value for China	6704.52	16203.13	18533.65	39441.87	53083.81
Average value for the West	4008	9281.1	10700.9	26209.9	37863.2
Average value for the East	10318.33	25130.42	28542.17	54620.17	72095.5
Average value for the Center	4882.22	11991.22	13892	33906.33	44646.67
Municipal level per capita GDP (yuan)
Average value for China	—	15412.76	—	38621.51	50961.96
Average value for the West	—	11035.33	—	28534.88	39143.08
Average value for the East	—	20536.40	—	47061.31	62022.64
Average value for the Center	—	12523.51	—	35639.56	46238.91
Permanent resident population of province (million people)
Average value for China	39.4587	41.48516	41.7816	43.23935	44.2219
Average value for the West	28.228	28.854	28.883	29.095	29.826
Average value for the East	41.9917	46.34083	47.0233	50.0758	51.4142
Average value for the Center	48.56	49.0456	49.1244	49.84	50.6278
Highway mileage of province (km)
Average value for China	39561.45	62275.55	111516	132464.1	147654.7
Average value for the West	36813.5	63930.9	104188.2	135690	155411.2
Average value for the East	36938.25	52609.33	90317.2	101470.7	113523.2
Average value for the Center	46112.33	73324.56	147923	170204.3	184545.2

Method

This study used a nation–region–province hierarchical structure for the country. Using the province as the underlying regional unit, both the CV and the Theil index were used to quantify national and regional differences, and the spatial correlation dimension was introduced to calculate the development of highway traffic for each province.

Coefficient of Variation

CV indicates the relative dispersion of data, which is calculated via dividing the sample standard deviation by the mean. According to the characteristics of CV, the per capita GDP, the density of the highway network, and the completion investment of the highway construction were selected for each province to measure spatial and temporal differences between China’s road traffic construction and its economic development at different scales. Difference indicators were calculated one by one between 1997 and 2015 and the equilibrium degree and the relationship between highway transportation development and economic development of China were dynamically investigated. A larger CV indicates increasing difference between provinces or regions.

CV is defined as a proportion

C_{v} = \frac{1}{\bar{y}} \sqrt{\sum_{i = 1}^{n} \frac{{(y_{i} - \bar{y})}^{2}}{n - 1}}

(1)

where $n$ represents the total number of units corresponding to a specific scale (nation scale, region scale, and province scale); $y_{i}$ represents the per capita GDP, the density of the highway network, or the completion of investment in highway construction of the $i$ unit; and $\bar{y}$ represents the average value of $y_{i}$ . The variation coefficient of per capita GDP ( $C_{v - per capita GDP}$ ), the density of the highway network ( $C_{v - density of highway}$ ), and the completion investment of the highway construction ( $C_{v - investment}$ ) at national scale and at three economic regional scales were calculated.

Theil Index

First order decomposition of the Theil index can be easily extended to an analysis of the between-region and within-region components of the national inequality. The overall national difference can be measured via the Theil index ( $T_{P}$ ) as

T_{P} = \sum_{i} \sum_{j} \frac{y_{ij}}{Y} \cdot \log [\frac{(\frac{y_{ij}}{Y})}{(\frac{p_{ij}}{P})}]

(2)

where $y_{ij}$ represents the highway mileage or GDP of province j in region i; $Y$ represents the total highway mileage or total GDP of all provinces ( $= \sum_{i} \sum_{j} y_{ij}$ ); $p_{ij}$ represents the permanent resident population of province j in region i; and $P$ represents the total permanent resident population of all provinces ( $= \sum_{i} \sum_{j} p_{ij}$ ).

If we define $T_{Pi}$ to measure between-province difference for region i, $T_{Pi}$ could be expressed as

T_{Pi} = \sum_{j} \frac{y_{ij}}{Y_{i}} \cdot \log [\frac{(\frac{y_{ij}}{Y_{i}})}{(\frac{p_{ij}}{P_{i}})}]

(3)

Then, the national Theil index $T_{P}$ of Equation 2 can be decomposed into

\begin{matrix} T_{P} = \sum_{i} \frac{Y_{i}}{Y} T_{Pi} + \sum_{i} \frac{Y_{i}}{Y} \log [\frac{(\frac{Y_{i}}{Y})}{(\frac{P_{i}}{P})}] \\ = \sum_{i} \frac{Y_{i}}{Y} T_{Pi} + T_{BR} \\ = T_{WR} + T_{BR} \end{matrix}

(4)

where $Y_{i}$ represents the total highway mileage or total GDP of region i ( $= \sum_{j} y_{ij}$ ); $P_{i}$ represents the permanent resident population of region i ( $= \sum_{j} p_{ij}$ ); $T_{BR}$ represents the difference between western, eastern, and central regions; $T_{WR}$ represents differences within-region. Equation 4 presents the ordinary one-stage Theil inequality decomposition. The total difference $T_{WR}$ can be expressed as the sum of a “within-region” difference term and a “between-region” term. Here, the within-region contribution is itself a weighted sum of the between-province difference of each region ( $T_{Pi}$ ).

Later, the Theil index of highway mileage was calculated as $T_{P - length of highway}$ and the GDP as $T_{P - GDP}$ , which provided directions and amplitudes of fluctuation of between-region and within-region differences of highway traffic development and economic development.

Space Relation Fractal Dimension

After analysis of CV and T_P, we obtained both the spatial and temporal differences of highway traffic development and economic development in China as a whole, and as segmented into different economic regions. The geographical location, resource endowment, and the spatial distribution of the inner cities and towns differed for each province. Next, the spatial correlation dimension indicated spatial connectivity of urban systems, and the development and connectivity indices of highway traffic were calculated for provinces; then, correlations with the economic differences of provinces were investigated.

The areal inhomogeneity of cities can be sufficiently characterized via the fractal correlation dimension D of the network ( 27 ). To obtain D of each province, the spatial correlation function $C_{r}$ must be considered, expressing the probability that two randomly selected cities of one province are within a distance below r units as

C_{r} = \frac{1}{N^{2}} \sum_{i, j = 1}^{N} θ (r - d_{ij}) (i \neq j)

(5)

where $N$ represents the number of cities in a province; $d_{ij}$ represents the spatial distance between two cities; $θ$ represents the Heaviside function, which assumes the value of 1, when the distance is lower than r or 0, and when the distance is larger, the value can be calculated via

θ (r - d_{ij}) = {\begin{matrix} 1 d_{ij} \leq r \\ 0 d_{ij} > r \end{matrix}

(6)

The spatial distribution of the city unit has fractal characteristics, and should therefore be scale invariant; then, $C_{r}$ grows like a power function ( 27 ), namely: $C_{r} \propto r^{D}$ , the deformation equation can be obtained as

{lnC}_{r} = Dlnr + const

(7)

where $r$ represents a given threshold, which belongs to the interval $[r_{\min}$ , $r_{\max}]$ , the value of which depends on the maximum and minimum distance of each province. For each province, $C_{r}$ was calculated by increasing the value of $r$ . Via linear regression, the correlation dimension D of the spatial distance can be determined, reflecting the inhomogeneity of the main cities in a province. In general, the range of D varies from 0 to 2 and when D approaches 0, the distribution of cities is highly concentrated; when D approaches 1, cities are distributed along a geographical line, and when D approaches 2, the spatial distribution of cities is very uniform. A smaller D value indicates higher concentration of the main cities in a province and a weaker interval resistance; in contrast, lower concentration indicates larger spacing resistance ( 28 ).

The correlation dimension of the actual traffic distance ( $D'$ ) can be calculated via changing $d_{ij}$ in Equation 5 to the actual traffic distance between both cities, and a closer $D'$ to $D$ indicates better connectivity between cities in the province.

The traffic connectivity index ρ is calculated via

ρ = \frac{D'}{D}

(8)

$ρ$ =1 indicates that the spatial accessibility between cities reached the ideal state. A smaller $ρ$ value indicates decreased accessibility between cities. If $ρ$ is above 1, the traffic network is inadequate.

Analysis and Result

General Analysis of Highway Development and Economic Development of China

The Chinese government invested massively into road construction to accelerate the expansion of the road network throughout counties and towns, to improve the overall quality of roads, and to increase the total mileage of expressways. Since 1984, the Chinese government has further increased its investment in roads infrustructure, particularly for high-quality roads such as highways that connect major industrial centers in coastal areas. During the 1990s, infrastructure investments became a national priority. During the early 2000s, road projects were an important part of China’s strategy for the development of the western and the central regions. In 2005, the Western Development Strategy and the Central Rise Policy were launched to improve transportation to the interior of China.

To contribute toward the building of a moderately prosperous society, the highway construction of China has currently entered a vital period. The aim of the “13th Five-Year Plan” was to connect all counties at least via secondary highways as well as to connect counties with freeways where topographical, technical, and economical conditions permit this, thus realizing an inter-village commuter bus in 2020. The completion of this highway construction investment from 1980 to 2016 in China is shown in Figure 2. This reflects that the policy investment effort of highway construction in China increased steadily and that the annual rate of investment growth levelled during recent years. The growth rates averaged 18.5%, 15.8%, and 7.2%, respectively, during 2002–2006, 2007–2011, and 2012–2016.

Figure 2.

Completion of highway construction in China over time.

Both the reform and introduction of the long-term implementation of public investment was concentrated in the coastal regions and in urban areas. Therefore, large regional variations exist within the density and quality of the highway infrastructure in China (Table 2). The western region is poorly connected compared to the central and eastern regions. In 2002, only 19 km of roads existed for every 100 km² of land in western China, compared to more than 58 km per 100 km² of land in the eastern region. Among all provinces, Tibet and Qinghai are particularly poorly equipped with road infrastructure with a road density of only 6 km and 10 km per 100 km² of land in 2015, respectively. Road quality is also worst in the western region and the construction of a freeway in Tibet began in 2014.

Table 2.

Comparison of the Highway Development of Different Regions

	Average completion of highway construction investmentbillion yuan				Density of highway networkKm/(100 km²)				Density of freeway networkKm/(100 km²)
Year	National	West	East	Center	National	West	East	Center	National	West	East	Center
1997	4.05	2.16	6.16	3.34	27.3	13.4	40.7	21.9	0.19	0.01	0.42	0.08
2000	7.47	5.56	9.49	6.9	31.8	15.5	48.2	24.8	0.53	0.11	1.12	0.22
2002	10.36	7.86	11.95	11.03	38.6	18.9	56.6	32.7	0.82	0.21	1.63	0.43
2003	11.98	8.62	13.38	13.86	39.7	19.4	58.3	33.6	0.97	0.28	1.90	0.51
2004	15.17	10.49	17.06	17.84	41.7	19.8	62.1	34.6	1.18	0.33	2.32	0.61
2009	31.19	25.1	33.06	35.46	80.8	47.1	104.7	78.3	1.36	0.39	2.64	0.76
2010	37.04	34.12	36.15	41.47	83.6	49.3	107.4	81.7	1.50	0.43	2.88	0.85
2011	40.63	38.73	39.27	44.58	85.4	51.0	109.3	83.4	1.68	0.54	3.09	1.05
2012	41.01	39.87	39.52	44.28	88.0	53.0	112.2	85.8	1.88	0.61	3.47	1.18
2013	44.17	45.49	40.74	47.28	90.2	54.7	115.2	87.5	2.06	0.70	3.79	1.27
2014	49.87	55.9	43.8	51.28	92.3	56.7	118.0	88.7	2.26	0.83	4.04	1.49
2015	53.27	59.51	46.6	55.23	94.9	59.5	120.3	90.9	2.46	0.93	4.26	1.74

In 1997, the investment ratio of the west to the east and to the center was 2:6:3. Completed investments for the center exceeded those of the east in 2003. Furthermore, in 2012, the west exceeded the east. In 2010, completed investments of the east were below the national average. Currently, China is vigorously promoting infrastructure equalization.

In China, most coastal provinces in the eastern region are economically rich, while most of the provincial areas of the center and the west are among the poorer provinces. Furthermore, the state of economic disparity between regions has continued, which is shown in Figure 1. The lack of transportation infrastructure has caused a development bottleneck, restricting the economic development of the western region as well as that of remote mountainous areas ( 3 , 21 ).

Difference Analysis Based on the CV

Considering the provincial administrative unit as the basic analytic unit, the national and regional $C_{v - density of highway}$ , $C_{v - per capita \ GDP}$ , $C_{v - investment}$ were calculated and the resulting trends are shown in Figure 3.

Figure 3.

CV of highway traffic development and economic development for different regions for (a) density of highway, (b) per capita GDP, and (c) investment.

Between 1997 and 2015, the development of highways in the western region was quite different, and the national difference was considerable, while the development of the center was relatively balanced. Its development and evolution can be divided into two stages: a strong fluctuation in the difference of highway development existed during 1997–2005, and the difference of highway development decreased and stabilized during 2006–2015. In 2006, the first year of the 11th Five-Year Plan, a highway expansion project and other projects were conducted as a result of which, the national and eastern difference of the highway network was noticeably reduced; however, the difference of western and central regions increased.

The regional economic disparity of China is constantly decreasing, especially among the eastern provinces; however, it still remains at a high level. During 2005–2006, investments were concentrated in the Henan and Hubei provinces of the central region, and as a consequence, the economic difference of the central region first increased, while it recently follows a downward trend; the production levels of western provinces are generally low and economic differences are small and fluctuate slightly.

According to the national $C_{v - investment}$ , the difference of the total social completed investment can be divided into two stages: 1997–2010 formed the great provincial difference in highway investment, as a result of concentrated investment, which was only conducted in some provinces; during 2011–2015, the investment difference remained stable. This trend of regional $C_{v - investment}$ reflects differences of the regional investment intensity.

The national $C_{v - density of highway}$ and $C_{v - per capita GDP}$ are positively correlated at the 0.01 level and the Pearson correlation coefficient reached 0.911. Over the course of highway construction, strengthening of a balanced development of highway transportation can reduce differences of economic development and promote a balanced development of the regional economy.

Difference Analysis Based on the Theil Index

Using the provincial administrative unit as the basic unit of analysis, the national and regional $T_{P - length of highway}$ and $T_{P - GDP}$ , were calculated between three economic regions, and within-region Thiel index. The obtained trends are shown in Figure 4.

Figure 4.

T_p of highway traffic development and economic development in different regions for (a) length of highway, (b) first order decomposition of national length of highway, (c) GDP, and (d) first order decomposition of national GDP.

Figure 4a provides the same conclusions as the analysis of variance coefficients. The difference of highway development and its evolution can be divided into two stages: 1997–2005 and 2006–2017. The difference of $T_{P - length of highway}$ changed relative to the population level; however, the sizes of provinces vary widely and with the growth of highway mileage in the western region, the difference of national highway mileage will continue to grow. The density of the highway network can represent the equalization level of highway development.

Since 1997, the difference of highway development in China has mainly manifested as an inter-regional difference, which has been decreasing during recent years, while between-region differences have gradually increased. In 2008, the total length of highways in the western region exceeded that of the eastern region for the first time and between-region differences gradually developed into the main cause, increasing annually until these differences finally steadied. The difference of China’s economic development can mainly be ascribed to between-region differences. It decreased annually and is currently equal to within-region. Investments carried out in the western and the central regions have achieved a reduction of this difference.

In the process of equilibrium development of the highway construction, the difference between developed and underdeveloped provinces of regions should be predominantly reduced.

Analysis of Highway Traffic Connectivity and Provincial Economic Difference

The connectivity of highway traffic was calculated via spatial correlation dimension, and the relationship between highway traffic connectivity and economical difference was verified in the provinces. The lower triangular matrixes are known for each province (except for four province-level municipalities) in which nondiagonal elements are equal to the spatial distance and the actual traffic distance between the corresponding vertices provided by main cities, while the diagonal elements all equal zero. This study used a step Δr of 20 km, according to both maximum and minimum of the spatial distance between the main cities in the province. Furthermore, the range of the r value in each province was adjusted, obtaining the series r and C(r); then, the scatter diagrams of ln(r) and lnC(r) were drawn, and the spatial correlation dimension D of each province was obtained via least square fit. D’ was obtained with the same method, simply by changing the space distance into the actual traffic distance. The calculated results are shown in Table 3.

Table 3.

Highway Traffic Connectivity and Economic Differences of Provinces

	Spatial correlation dimension			CV
	D(R²)	$D^{'}$ (R²)	$ρ$	2005	2011	2015
West				West
Sichuan	1.4579 (0.93)	1.3055 (0.92)	0.8955	0.5463	0.4203	0.4274
Guizhou	1.5707 (0.88)	1.449 (0.92)	0.9225	0.6222	0.4470	0.4271
Yunnan	1.4649 (0.95)	1.3765 (0.96)	0.9397	0.7360	0.6399	0.5838
Tibet	0.9426 (0.97)	1.0404 (0.97)	1.1038	—	—	—
Shaanxi	1.6728 (0.90)	1.5097 (0.93)	0.9025	0.4945	0.5454	0.3957
Gansu	1.2062 (0.88)	0.8939 (0.89)	0.7411	1.0026	0.9260	0.5906
Qinghai	1.5035 (0.91)	1.3304 (0.91)	0.8849	—	—	—
Ningxia	1.2605 (0.97)	1.0858 (0.98)	0.8614	0.6704	0.6045	0.5527
Xinjiang	1.2218 (0.95)	1.0601 (0.94)	0.8677	0.7817	0.5949	0.3903
East				East
Hebei	1.3996 (0.94)	1.2611 (0.95)	0.9010	0.3659	0.4251	0.3812
Liaoning	1.6425 (0.94)	1.5316 (0.92)	0.9325	0.5471	0.4313	0.4549
Jiangsu	1.2591 (0.84)	1.2581 (0.85)	0.9992	0.6853	0.4337	0.3694
Zhejiang	1.7356 (0.95)	1.6481 (0.96)	0.9496	0.3941	0.3083	0.2914
Fujian	1.6007 (0.92)	1.4298 (0.95)	0.8932	0.5697	0.2647	0.1963
Shandong	1.7395 (0.93)	1.5312 (0.92)	0.8803	0.6190	0.5027	0.4843
Guangdong	1.2406 (0.91)	1.1656 (0.93)	0.9395	0.8416	0.6792	0.6762
Guangxi	1.5321 (0.97)	1.4375 (0.98)	0.9383	0.3822	0.3753	0.4299
Hainan	1.4188 (0.94)	1.4098 (0.93)	0.9937	0.1290	0.1701	0.07582
Center				Center
Shanxi	1.5295 (0.91)	1.3604 (0.92)	0.8894	0.4110	0.3304	0.3701
Inner Mongolia	1.346 (0.94)	1.0102 (0.92)	0.7505	0.6409	0.6250	0.6013
Jilin	1.393 (0.94)	1.354 (0.93)	0.9720	0.3678	0.2365	0.2390
Heilongjiang	1.3368 (0.91)	1.1735 (0.91)	0.8778	0.9273	0.8904	0.6725
Anhui	1.5565 (0.89)	1.5014 (0.92)	0.9646	0.6441	0.5945	0.4543
Jiangxi	1.629 (0.98)	1.5483 (0.96)	0.9505	0.4817	0.5411	0.4676
Henan	1.766 (0.91)	1.6295 (0.91)	0.9227	0.4292	0.3761	0.3462
Hubei	1.4029 (0.97)	1.2725 (0.95)	0.9070	0.5175	0.4606	0.4707
Hunan	1.7404 (0.93)	1.4231 (0.94)	0.8177	0.4645	0.5714	0.5637

Economic differences within each province are represented by the CV of per capita GDP of prefecture-level cities. The calculated results are also shown in Table 3.

The D values indicate the distribution and evenness of the main cities in the province. The distribution characteristics of the main cities in Tibet are most apparent, and the main cities in this province are distributed along a curve. However, only Grade I and II roads are available for communication and exchange between cities in the province or between cities of other provinces that lack a freeway network. The D values of Gansu, Ningxia, Xinjiang, Jiangsu, Guangdong, Inner Mongolia, and Heilongjiang are all below 1.35. Due to their long and narrow geographical boundaries, their main cities are distributed along geographical lines. The main cities in the remainder of the province follow a uniform spatial distribution. The ρ values of Jiangsu and Hainan were close to 1, indicating superior highway traffic development. The ρ-value of Tibet was above 1, while for Gansu and Inner Mongolia the ρ-value was below 0.8, indicating poor highway traffic connectivity. Due to the limitation of geographical environment in the provinces of western and central China, the distances between main cities are large, and the connectivity resistance is high; consequently, several main cities are not covered, leading to an incomplete highway network. For example, the freeways of Sichuan are mainly concentrated in the east, near Chongqing, leading to a lack of high-grade road connections in the western region. Figure 5a shows that the freeway network of the study area features an apparent spatial agglomeration, which is concentrated in four province-level municipalities (Beijing, Tianjin, Shanghai, and Chongqing) and in Guangzhou City. However, the vast regions of Xinjiang, Gansu, Qinghai, Tibet, and Heilongjiang lack a developed highway network. The connectivity index ρ distribution of study area is shown in Figure 5b; here, the ρ-value of Tibet is equal to that of Gansu.

Figure 5.

Development of highway traffic in provinces showing (a) freeway network and main cities, and (b) connectivity of provinces.

CV-values of GDP per capita decreased in most of the provinces; however, some first increased and then decreased, such as in Hunan, Jiangxi, Shaanxi, and Hebei. The CV of GDP per capita has retained a high level in Yunnan, Gansu, Ningxia, Guangdong, Inner Mongolia, and Heilongjiang, while economic differences were large. The highway traffic network in these 10 provinces shows poor connectivity and several remote main cities were not connected via high-grade highways. Traffic connectivity is the main reason for uneven economic development in the province.

Statistics of GDP per capita are incomplete for the prefecture level cities of Tibet and Qinghai, where the CV has not been calculated yet. However, judging from already existing data, Tibet and Qinghai are economically underdeveloped regions, and economic differences in the province are apparent. We conducted a correlation analysis between ρ-value and CV for 2011 (except for Tibet and Qinghai). The results show that they are significantly negatively correlated at the 0.01 level with a correlation coefficient of −0.581. Thus, the connectivity of highway traffic in the provinces is significantly related to the economic differences of the provinces.

Conclusion and Discussion

During the studied period, the differences between economic development and highway traffic development were strongly related. China is currently improving the disequilibrium of highway traffic development introduced by the previous biased policy. This study calculated and decomposed the difference of highway traffic development and economic development of China by using the CV and the Theil index. For future highway construction policies, the obtained findings offer clues regarding the suitability of implementing rebalancing policies in terms of national or regional design.

The investment difference can be divided into two stages: during 1997–2010, the government focused on investing into central or western provinces and consequently the variation of investment was large and unstable; during 2011–2016, the investment difference remained between 0.62–0.65. The difference of highway traffic development was also divided into two stages: 1997–2006 and 2007–2015. In approximately 2006, the national and eastern difference was significantly reduced, while the difference between western and central regions firstly increased and then decreased.

With the implementation of highway infrastructure construction projects, the national economic difference decreased and the resulting reaction of the three economic regions differed. The economic difference of the eastern region continued to decrease, while that of the central region first increased and then decreased (following an inverse U-shape), and that of the western region generally remained at a low level of productivity, where economic differences only fluctuated slightly.

For a long time, within-region differences contributed predominantly to the differences of highway traffic development. With the increasing highway mileage in the western region, the difference between-region has been increasing as well, which is an inevitable phenomenon and will eventually reach a steady state. The implementation of a novel investment policy has played a significant role in reducing the once dominant between-region economic differences.

The connectivity of highway traffic strongly correlates with the economic difference of a province, and improved connectivity leads to smaller economic difference in that province. However, traffic connectivity of provinces in the central and western regions remains poor. Due to many restrictions of the geographical environment, the distances between cities are large, and some main cities are not connected via high-grade highways.

Intercity railways are currently rapidly developing; however, highway transportation can provide a door-to-door service, which enables a more vigorous economic development for the region and has a significant potential to promote a balanced economic development. In future, the equalized development of highway traffic in China should first reduce the existing differences between developed and underdeveloped areas of each region. Then, high-grade highways should be built deep into remote areas.

The change of traffic mileage was less sensitive to the development of a highway network, which makes it more advantageous to study intercity connectivity or accessibility in response to travel time. With the recent development of map services (e.g. Baidu map and API interface), data support has been provided and as a consequence, studying the relationship between multimodal transportation development and economic or industrial development at different scales became convenient.

Footnotes

Acknowledgements

This study was supported by the projects of the National Natural Science Foundation of China (71774118, 91546115).

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: Ye Li, Jing Fan, Haopeng Deng; data collection: Jing Fan; analysis and interpretation of results: Jing Fan; draft manuscript preparation: Ye Li, Jing Fan, Haopeng Deng. All authors reviewed the results and approved the final version of the manuscript.

The Standing Committee on Transportation and Economic Development (ADD10) peer-reviewed this paper (18-02800).

References

Fan

Zhang

Growth, Inequality, and Poverty in Rural China: The Role of Public Investments, Vol. 125. International Food Policy Research Institute, Washington, D.C., 2002.

Wang

H. Z.

Transportation and Regional Economic Discrepancy-Empirical Analysis of Railways in China. Journal of Shanxi Finance and Economics University, Vol. 33, No. 2, 2011, pp. 61–68.

Fan

Chan-Kang

Road Development, Economic Growth, and Poverty Reduction in China, Vol. 12. International Food Policy Research Institute, Washington, D.C., 2005.

Aschauer

D. A.

Is Public Expenditure Productive?

Journal of Monetary Economics, Vol. 23, No. 2, 1989, pp. 177–200.

Arbués

José

F. B.

Mayor

The Spatial Productivity of Transportation Infrastructure. Transportation Research Part A: Policy and Practice, Vol. 75, 2015, pp. 166–177.

Lakshmanan

T. R.

The Broader Economic Consequences of Transport Infrastructure Investments. Journal of Transport Geography, Vol. 19, No. 1, 2011, pp. 1–12.

Ozbay

Ozmen-Ertekin

Joseph

Contribution of Transportation Investments to County Output. Transport Policy, Vol. 14, No. 4, 2007, pp. 317–329.

Cantos

Gumbau-Albert

Maudos

Transport Infrastructures, Spillover Effects and Regional Growth: Evidence of the Spanish Case. Transport Reviews, Vol. 25, No. 1, 2005, pp. 25–50.

Boarnet

M. G.

The Direct and Indirect Economic Effects of Transportation Infrastructure. University of California Transportation Center, 1996.

10.

Mikesell

J. L.

Wang

J. Q.

Zhao

Z. J.

Impact of Transportation Investment on Economic Growth in China. Transportation Research Record Journal of the Transportation Research Board, 2016. 2531: 9–16.

11.

Berechman

Ozmen

Ozbay

Empirical Analysis of Transportation Investment and Economic Development at State, County and Municipality Levels. Transportation, Vol. 33, No. 6, 2006, pp. 537–551.

12.

Chen

Haynes

K. E.

Transportation Infrastructure and Economic Growth in China: A Meta-Analysis. In Socioeconomic Environmental Policies and Evaluations in Regional Science ( Shibusawa

Sakurai

Mizunoya

Uchida

, eds.), Springer, Singapore, 2017.

13.

Pereira

A. M.

Andraz

J. M.

Public Investment in Transportation Infrastructure and Industry Performance in Portugal. Journal of Economic Development, Vol. 32, No. 1, 2007, p. 1.

14.

Jiang

Zhang

Qin

Shao

Multimodal Transportation Infrastructure Investment and Regional Economic Development: A Structural Equation Modeling Empirical Analysis in China from 1986 to 2011. Transport Policy, Vol. 54, 2017, pp. 43–52.

15.

Straszheim

M. R.

Researching the Role of Transportation in Regional Development. Land Economics, Vol. 48, No. 3, 1972, pp. 212–219.

16.

Vickerman

Spiekermann

Wegener

Accessibility and Economic Development in Europe. Regional Studies, Vol. 33, No. 1, 1999, pp. 1–15.

17.

De Jong

Storm

The Growth Impact of Transport Infrastructure Investment: A Regional Analysis for China (1978–2008). Policy and Society, Vol. 31, No. 1, 2012, pp. 25–38.

18.

Zhang

X. L.

Regional Comparative Analysis on the Relationship between Transport Infrastructure and Economic Growth in China. Journal of Finance and Economics, Vol. 33, No. 8, 2007, pp. 51–63.

19.

Wang

J. W.

Mao

Spatial Disparities of the Distribution of Highway Network in Northwest China based on Exploratory Spatial Data Analysis. Arid Land Geography, Vol. 36, No. 2, 2013, pp. 329–336.

20.

Cao

Zhang

Xue

Wang

The Changes in Disparity of Urban Transportation Development Level Rank in China. Acta Geographica Sinica-Chinese Edition, Vol. 62, No. 10, 2007, pp. 1034–1040.

21.

Demurger

Infrastructure Development and Economic Growth: An Explanation for Regional Disparities in China?

Journal of Comparative Economics, Vol. 29, No. 1, 2001, pp. 95–117.

22.

S. M.

Shum

Y. M.

Impacts of the National Trunk Highway System on Accessibility in China. Journal of Transport Geography, Vol. 9, No. 1, 2001, pp. 39–48.

23.

Chen

P. Y.

Zhu

X. G.

Regional Inequalities in China at Different Scales. Acta Geographica Sinica, Vol. 67, No. 8, 2012, pp. 1085–1097.

24.

Bedeian

A. G.

Mossholder

K. W.

On the Use of the Coefficient of Variation as a Measure of Diversity. Organizational Research Methods, Vol. 3, No. 3, 2000, pp. 285–297.

25.

Bourguignon

Decomposable Income Inequality Measures. Econometrica, Vol. 47, No. 4, 1979, pp. 901–920.

26.

Duro

J. A.

Teixidó-Figueras

Ecological Footprint Inequality across Countries: The Role of Environment Intensity, Income and Interaction Effects. Ecological Economics, Vol. 93, No. 3, 2013, pp. 34–41.

27.

Capecchi

Crisci

Melani

Morabito

Politi

Fractal Characterization of Rain-gauge Networks and Precipitations: An Application in Central Italy. Theoretical & Applied Climatology, Vol. 107, No. 3–4, 2012, pp. 541–546.

28.

Liu

J. S.

Chen

Y. G.

A Study on Fractal Dimensions of Spatial Structure of Networks and the Methods of Their Determination. Acta Geographica Sinica-Chinese Edition, Vol. 54, 1999, pp. 471–478.