You are on page 1of 25

Social Sciences in China

ISSN: 0252-9203 (Print) 1940-5952 (Online) Journal homepage: http://www.tandfonline.com/loi/rssc20

Has Transport Infrastructure Promoted Regional


Economic Growth?— With an Analysis of the
Spatial Spillover Effects of Transport Infrastructure

Zhang Xueliang

To cite this article: Zhang Xueliang (2013) Has Transport Infrastructure Promoted Regional
Economic Growth?— With an Analysis of the Spatial Spillover Effects of Transport Infrastructure,
Social Sciences in China, 34:2, 24-47, DOI: 10.1080/02529203.2013.787222

To link to this article: https://doi.org/10.1080/02529203.2013.787222

Published online: 29 Apr 2013.

Submit your article to this journal

Article views: 399

View related articles

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=rssc20
Social Sciences in China, 2013
Vol. 34, No. 2, 24-47, http://dx.doi.org/10.1080/02529203.2013.787222

Has Transport Infrastructure Promoted Regional Economic Growth?—


With an Analysis of the Spatial Spillover Effects of Transport
Infrastructure*
Zhang Xueliang

Institute of Finance and Economics, Shanghai University of Finance and Economics

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

* This study is financed by the Youth Project of the National Social Science Foundation “Studies
on the Spatial Spillover Effects of Transport Infrastructure on Chinese Regional Economic Growth”
(No. 70803030), the Shanghai “Shuguang” Project of 2011 (No. 11SG36), and the Key Scientific
Research Innovation Project of the Shanghai Education Commission (No. 10ZS50). The author hereby
expresses his thanks to Professor Zhang Jun from Fudan University and Professor Ai Chunrong from the
University of Florida, USA, for their valuable suggestions.
© Social Sciences in China Press
Zhang Xueliang 25

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.

Keywords: transport infrastructure, spatial spillover effects, regional economic growth,


spatial econometric model

Investment in transport infrastructure with its classic externalities is always an important


government responsibility as well as an important means by which the government can
regulate the economy and promote economic growth. Take China for example: in order to
rise to the challenge of the international financial crisis and maintain steady and relatively
rapid economic development, the Chinese central government introduced a four-trillion
yuan economic stimulus package at the end of 2008, with more than half of the money being
invested in railway, highway, airport and other major transport infrastructure construction.
The opening to traffic of the whole Beijing-Shanghai high-speed railway in 2011 was a
landmark event of this round of investment in infrastructure. It greatly reduces the time
needed for the ride from the Yangtze Delta city cluster to the Bohai Rim city cluster and has
produced clear regionally coordinated urbanization effects. Such a large-scale investment
in transport infrastructure will have far-reaching significance for regional economic growth.
What concerns us here is the efficiency of public capital investments, including the investment
in transport infrastructure, against the background of economic globalization and regionally
coordinated urbanization. Has transport infrastructure promoted regional economic growth
in China? How can the effectiveness of China’s regional transport policy be improved in the
context of the national strategy of narrowing the gap between regions and achieving balanced
regional development? The results of this study show that in the analysis of the total effect
of transport infrastructure on economic growth, consideration should be given to spatial
spillover effects, otherwise the role of transport infrastructure would be overestimated. In
China, for a specific region, transport infrastructure in other regions has mainly positive
spillover effects on economic growth and we should not attribute the widening income gap
between regions mainly to the construction of transport infrastructure. However, our research
has also discovered evidence of the negative spillover effects of transport infrastructure on
regional economic growth in the context of the predominantly one-way flow of labor from
the backward areas in the central and western regions to the eastern coastal areas, a discovery
that has important practical significance for an objective appraisal of the role of transport
infrastructure in regional economic growth.

I. Literature Review and Analysis

The relationship between transport infrastructure and economic growth is a question which
26 Social Sciences in China

economists have taken very seriously. Especially in the 1940s, Paul Rosenstein-Rodan,1
Ragnar Nurkse,2 W.W. Rostow3 and other development economists began to put forward a
number of insightful ideas on the relationship between infrastructure, including transport
infrastructure, and economic growth, ideas which were widely used to guide the practice
of developing countries. Rodin was the first to propose the “big push” theory, believing
that transport and other infrastructure constituted social overhead capital and must be
given priority in development. Rostow also saw transport and other infrastructure as social
overhead capital and regarded them as a vital precondition for “economic take-off.” Nurkse
developed Rodin’s theory and held the view that investment in transport infrastructure is the
responsibility of the government because private enterprise seldom has an incentive to invest
in transport infrastructure, with its indivisible initial investment and strong externalities.
The theories of these development economists were once extensively practiced in many
developing countries, but did not achieve the desired results. Transportation and other
infrastructure as social overhead capital did not promote socio-economic development in the
developing world as significantly as expected. With the widespread application of empirical
analysis and in light of the decrease in public capital investment in the United States in
the late 1960s followed by the decline in productivity that took place around 1973, David
Aschauer,4 using a neoclassical economic growth model, carried out economic research
that involved putting together the data on decreased investment in transport and other
infrastructure and subsequent productivity, and came to the conclusion that transport and other
infrastructure had an important role in economic growth. He also calculated that the output
elasticity of transport and other infrastructure was 0.39. Aschauer’s study sparked an upsurge
of research among scholars on the impact of transport infrastructure capital on total output
and productivity growth. Some studies using time series data showed that the elasticity of
transport infrastructure in relation to economic growth was between 0.27 and 0.58. Bonaglia
et al.5 questioned this high output elasticity and thought there might be spurious correlations
among the variables of the time series data. He also pointed out that the discrepancy between
different cross-sectional data in the model was not taken into account. For this reason, some
scholars turned to the use of provincial (state) panel data in their analysis. Early use of
panel data led to smaller estimated values of the output elasticity of transport infrastructure
than those calculated by Aschauer et al. using time series data; some studies even came to
the conclusion that transport infrastructure did not have any significant effect on economic
growth, a far cry from the common understanding of its positive effects on economic growth.

1 Paul Rosenstein-Rodan, “Problems of Industrialization of Eastern and South-Eastern Europe,” pp.


202-211.
2 Ragnar Nurkse, Problems of Capital Formation in Underdeveloped Countries.
3 W.W. Rostow, The Stages of Economic Growth: A Non-Communist Manifesto.
4 David Alan Aschauer, “Is Public Expenditure Productive?”, pp. 177-200.
5 F. Bonaglia, E. La Ferrara and M. Marcellino, “Public Capital and Economic Performance: Evidence
from Italy.”
Zhang Xueliang 27

There may be two reasons for the great difference between the conclusions reached by the
use of time series data and the use of panel data. First, the production functions used by the
two approaches contain different data structures and variable structures. If these structures
are not valid, there will be serious problems with the estimations. In addition, in most
cases the production functions used by both approaches take account only of the effects on
economic growth of variables such as labor and transport infrastructure, but do not analyze
the synergistic effects of new economic growth and economic geography factors on economic
growth. Second, neither approach takes account in its modeling of the spatial spillover effects
of transport infrastructure on economic growth. Most recent research papers are concerned
with whether transport infrastructure has any production effects, but give little attention to the
fact that transport infrastructure may shift local economic activities to other regions (Boarnet,6
Cantos et al.7). Transport infrastructure has network properties: it integrates the economic
activities of different regions and, through the diffusion effect, makes regions with faster
economic growth carry with them those regions that have slower economic growth, thus
demonstrating a positive spatial spillover effect. On the other hand, transport infrastructure
may produce a negative spatial spillover effect; that is, through its agglomerative effects, it
facilitates the movement of the factors of production to economically developed areas. In this
case, one region’s economic growth may be achieved at the expense of another region. The
problem ignored by a production function model that uses panel data to study the relationship
between transport infrastructure and economic growth is that the model fails to take into
account of the impact on economic growth of the spatial flow of capital, labor and other
production factors between regions via transport, thus giving no consideration to the spatial
spillover effects of transport infrastructure.
Of course, the models in the above-mentioned literature are mostly constructed within
the framework of neoclassical economic growth theory. In the absence of the new economic
growth and new economic geography variables, analysis of the role of transport infrastructure
in economic growth is likely to go astray. The new economic growth model focuses on
new economic growth factors such as knowledge accumulation and human capital. Over
recent years, spatial economic theory with increasing returns to scale, transportation costs
and incomplete competition as its cornerstone has developed rapidly, and a large group of
economists represented by Paul Krugman, who won the Nobel Prize in Economics in 2008,
have tried to introduce the spatial dimension into international economics and research on
industrial agglomeration.8 In empirical studies, the development of spatial statistics and spatial
econometrics has provided spatial economics with a very good empirical analytical tool while

6 M.G. Boarnet, “Spillover and the Locational Effects of Public Infrastructure,” pp. 381-400.
7 Pedro Cantos, Mercedes Gumbau-Albert and Joaquín Maudos, “Transport Infrastructures, Spillover
Effects and Regional Growth: Evidence of the Spanish Case,” pp. 25-50.
8 M. Fujita, P. Krugman and A.J. Venables, The Spatial Economy: Cities, Regions, and International
Trade.
28 Social Sciences in China

broadening traditional econometric theoretical methods.9 It may be said that the deepening of
research on economic growth theory, especially the development of spatial economic theory
and empirical analysis, has afforded new ideas and approaches for study of the relationship
between transport infrastructure and economic growth.
Since reform and opening up, China has witnessed a rapid increase in investment in
transport infrastructure. In the period from 1979 to 2009, China’s investment in transport
totaled 11 trillion yuan, with an average annual growth of 20.87 percent. The proportion of
transport investment in GDP has risen from less than 1.3 percent in 1953 to 6.6 percent in
recent years. In terms of regions, in 2009 transport infrastructure investment in the eastern,
central and western regions was 17.79, 26.09 and 35.19 times respectively what it was in
1993, with the increase in the central and western regions being much higher than in the
eastern region.10 Transport infrastructure, long seen as a bottleneck holding back regional
economic development, has gone ahead in leaps and bounds in recent years, but has also
shown features of unbalanced regional development in geographical space. Research is being
undertaken in this field; for example, Yu Yongjun and Lu Yulin11 constructed a theoretical
model on the basis of a rational ratio of transport investment to other productive investment
in order to probe into the relationship between transport investment and economic growth,
and Gao Feng12 and Zhang Xueliang13 used the panel data analysis method to estimate the
contribution of transport infrastructure investment to economic growth and the differences
in this respect between the eastern, central and western regions. Further efforts are needed to
study the role of transport infrastructure in regional economic growth from the perspective of
spatial spillovers within the framework of the synergistic effects of multidimensional factors.
China is a developing country. On the one hand, the promotion of economic growth
through transport infrastructure must be achieved through the synergistic effects of factors
such as the improvement of the institutional environment, the upgrading of human capital
and the raising of scientific, educational and technological levels, so the analysis of economic
growth must include a comprehensive analysis of the impact of multidimensional factors,
including new economic geography and new economic growth factors. On the other hand,
the spatial spillover effects of transport infrastructure on economic growth also need to be
examined. This study will use cutting-edge spatial econometric analytical techniques to

9 David W.S. Wong and Jay Lee, Statistical Analysis of Geographic Information with ArcView GIS
and ArcGIS, p. 2
10 The data come from the Economic Space Data Center of the Yangtze River Delta City Cluster,
Shanghai University of Finance and Economics. The Center obtained these data through calculations and
organization of material in the China Statistical Yearbook for various years compiled by the National
Bureau of Statistics of the People’s Republic of China and published by China Statistics Press.
11 Yu Yongjun and Lu Yuqi, “A Study of the Relationship between Transport Investment and
Economic Development and Methods of Evaluating Its Regional Effects.”
12 Gao Feng, Transport Infrastructure Investment and Economic Growth, p. 168.
13 Zhang Xueliang, “A Comparative Regional Analysis of Transport Infrastructure and Economic
Growth in China.”
Zhang Xueliang 29

study the spatial spillover effects of transport infrastructure on regional economic growth in
China in an economic growth model that gives comprehensive consideration to the spatial
synergistic effects of multidimensional factors, with a view to arriving at some new findings
in the course of remedying some of the defects of domestic research in this field. We will first
construct a basic model of the spatial spillover effects of transport infrastructure on regional
economic growth, and then, in the course of empirical analysis, construct an economic growth
model of the spatial synergistic effects of multidimensional factors. The two models use
provincial panel data respectively from 1993 to 2009 and from 2000 to 2009 and, employing
spatial econometric analytical techniques, analyze the spatial spillover effects of transport
infrastructure on economic growth.

II. The Basic Model

Traditional economic theory generally assumes that the flow of production factors in
geographical space is instantaneously completed, so the transportation of production factors is
cost-free and what we refer to as spatial spillover effects on regional economic development
cannot possibly exist. However, our empirical observations tell us that the transportation cost
of production factors and commodities can by no means be zero. Transport infrastructure,
including roads, waterways and railway and air transport, has typical network properties
and externalities and its impact on output growth is not limited to the places it passes
through but extends to adjacent regions. Besides the externalities possessed by such general
infrastructure as social public goods, transport infrastructure also has regional externalities.
The mechanism by which it exerts its spatial spillover effects is as follows: on the one hand,
transport infrastructure has network properties: it integrates the economic activities of
various regions, lowers transportation costs for businesses and residents and, through the
diffusion effect, makes regions with faster economic growth carry with them those with a
slower economic growth, thus showing positive spatial spillover effects. On the other hand,
transport infrastructure changes the accessibility and attractiveness of the region where it is
situated, improves that region’s geographical advantages and speeds up the flow of production
factors. Especially for economically developed regions with their first-mover advantages of
long-term accumulation of powerful scientific and technological strength, good institutional
environments, abundant capital and vast consumer markets, transport infrastructure expansion
will further enhance their competitive advantage but may also lead to economic recession
in other regions; that is, it may produce negative spillover effects on economic growth in
these regions, especially in economically backward regions. Therefore, in analyzing the total
effect of transport infrastructure on regional economic growth, we need to take account of its
spatial spillover effects, otherwise we may overestimate or underestimate the role of transport
infrastructure in regional economic growth.
The fundamental reason that transport infrastructure has spatial spillover effects on regional
30 Social Sciences in China

economic growth is its regional externalities, which are, in turn, closely linked with the
agglomeration and diffusion of the various factors of production in different regions. Regional
economic theory tells us that in the process of spatial agglomeration and factor diffusion,
transport infrastructure is the precondition for spatial agglomeration in the advanced regions
because a good transportation network helps bring various factors of production into the first-
mover regions; at the same time, it is also a condition of the spatial diffusion of advanced
regions because good transport infrastructure is a prerequisite for backward areas to attract
factor movement and industrial gradient transfer from the advanced regions. When the factors
of production flow mainly from backward to developed areas through a convenient transport
infrastructure and collect in the developed areas, transport infrastructure will have a negative
spatial spillover effect on economic growth in the backward regions, but when the spatial
mobility of production factors that is effected through transport infrastructure is expressed
mainly in spatial diffusion to the backward regions, transport infrastructure will have a
positive spatial spillover effect on their economic growth.
In our empirical analysis, we have used spatial econometric techniques to construct a
production function model that includes local transport infrastructure as well as transport
infrastructure in other relevant regions, and carried out an empirical analysis of the spatial
spillover effects of transport infrastructure on regional economic growth.14
Y = Af ( Kc, Kt , Kg , OKt , L, X ) (1)
in which Y stands for total output, A for technological progress, Kg for capital stock in other
parts of the public sector excluding transport infrastructure, Kt for capital stock in transport
infrastructure, L for labor input, and Kc for capital stock in the private sector. X represents
the vector composed of all other factors that impact on the total output, including new
economic growth and new economic geography factors such as human capital accumulation,
globalization, economic policy, and industrial agglomeration, which we will discuss in detail
in Part III of this article.
Equation (1) also includes another variable—transport infrastructure capital stock in other
relevant regions OKt; the spatial spillover effects of transport infrastructure on regional
economic growth, are measured by this variable. This article brings in spatial weight matrices
in spatial statistics and spatial econometrics to construct the variable OKt in Equation (1). The
constructed variable OKt can reflect not only the role of the transport infrastructure of other
relevant regions in the economic growth of a particular region, but also a variety of regional
economic ties. In this article, the specific structure of OKt is as follows:
N
OKti = ∑ wij Kt j (2)
j =1

14 A classic spatial econometric model consists of a spatial lag model (SLM) and a spatial error model
(SEM), focusing on consideration of the spatial spillover effects of the explained variables. Besides
the spatial spillover effects of the explanatory “transport infrastructure” variable, this study will also
consider the spatial spillover effects of the explained variables in the empirical research section.
Zhang Xueliang 31

in which N represents the number of regions adjacent to region i in the spatial weight matrix
and wij stands for the value of an element in the spatial weight matrix.
The spatial weight matrix introduced by spatial econometrics is one of its important
differences from the traditional econometrics. The binary symmetry spatial weight matrix Wnxn
represents n positions of spatial proximity relations, and the value of an element in a matrix can
be measured by the adjacency or distance standard. There are a variety of rules for constructing
spatial weight matrices. In practical applications, scholars construct appropriate spatial weight
matrix units according to the specific forms of spatial association corresponding to their
research objectives and issues. With reference to the spatial weight matrix construction methods
summarized by Bavaud,15 this study constructs the following four types of spatial weight matrix.
(1) The simplest binary 0-1 spatial weight matrix Wcont, as shown by Equation (3). If the
two regions are contiguous, the corresponding weight element value is 1; if they are not, the
corresponding weight element value is 0. Finally we standardize the matrix, making sure that
the sum of the elements in every row is equal to 1.
1 when regioni iand
Whenregion regionj jare
andregion arecontiguous
contiguous
wij =  (3)
0 when i = j or not contiguous
when i = j or not contiguous
(2) The spatial weight matrix that reflects geographical distance Wnet. Here the method for
setting up the weight element is as follows:
N ij
wij =
∑N ij
(4)
j

in which Nij stands for the number of trunk transport infrastructures (state highways) that
connect the two contiguous regions i and j, and j for the number of regions contiguous to
region i. If there is no main line of communication between the two contiguous regions i and
j, wij will be equal to 0. On the basis of data on state highway connections in the provinces,
municipalities and autonomous regions of the mainland of China in 1999, we created spatial
weight matrices for each year and lastly standardized them, ensuring that the sum of the
elements in every row was equal to 1. It should be noted that the construction of transport
infrastructure integrates all regions linked up by the transport network and economic
development in one region will promote economic development in the surrounding regions.
Therefore, the transport network spatial weight matrix can be used to analyze the positive
spatial spillover effects of transport infrastructure on economic growth.
(3) The population density spatial weight matrix Wperpop and the per capita GDP spatial
weight matrix Wpergdp, both reflecting economic distance. The computation formula is:
1/ X i − X j
wij = (5)
∑1/ X
j
i −Xj

15 Francois Bavaud, “Models for Spatial Weights: A Systematic Look,” pp. 153-171.
32 Social Sciences in China

in which Xi in Wperpop represents the population density expressed as number of people per
square kilometer in region i while Xi in Wpergdp represents the per capita GDP expressed as
GDP per ten thousand people in region i. Lastly we standardize them, ensuring that the sum
of the elements in every row is equal to 1. Here too it should be noted that the improvement
of transport infrastructure in one region will improve that region’s geographical advantage,
but an increase in its output may mean a decrease in the output of other regions. Since regions
with similar population density and per capita GDP generally compete with one another in
attracting the flow of production factors, the two types of matrices may be used to analyze the
negative spatial spillover effects of transport infrastructure on economic growth.

III. Extension of the Model and Definition of the Variables

Factors that affect regional economic growth are very complex and economic growth may be
the result of the spatial synergistic effects of multidimensional factors. Transport infrastructure
is an indispensable but not a sufficient condition for regional economic growth. 16 For
this reason, in this part of the article we introduce new economic growth, new economic
geography and other factors into our empirical study and construct an economic growth model
of the spatial synergistic effects of multidimensional factors. At the same time, we also take
note of the fact that due to the flow of production factors and the impact of the agglomeration
and demonstration effects of economic growth, economic growth in different regions also
shows spatial dependence and produces its own spatial spillover effects. We therefore also test
the spillover effects of the economic growth itself. The expanded model is as follows:
Y =Af ( ρWY, Kc, Kt, Kg, OKt, L, X ) (6)
in which Y represents actual GDP allowing for price factors for different years and different
regions, ρWY stands for the spatial lag effect used in analyzing regional economic growth,
and ρ for the spatial lag coefficient, which reflects the spatial dependency between sample
observations, i.e., whether the impact of the neighboring province’s observation value Y on a
particular province’s economic growth is positive or negative and the extent to which this is so.
Some variables are explained below.
(1) Measurement of various types of capital stock. The most commonly used method is
the perpetual inventory method. Its application mainly involves calculation of base period
capital stock, construction of the price indices of fixed assets investment, the depreciation
rate and selection of current investment index. With reference to the practice of Zhang et

16 Banister and Berechman discuss the three conditions under which transport infrastructure plays
a role in economic growth in developed countries: economic externalities, transport infrastructure
investment and the political and policy situation. They point out that only when the three conditions are
met simultaneously will it be possible to achieve economic development and that the existence of one
condition alone has almost no impact on economic development. Cf. D. Banister and Y. Berechman,
“Transport Investment and the Promotion of Economic Growth,” pp. 209-218.
Zhang Xueliang 33

al.,17 if we use δ to represent the rate of depreciation, we get the capital stock of industry j in
province i in the year t by the following equation:
Kijt=Kijt−1(1−δijt)+Iijt, t =2, …, 17 (7)
(a) Determination of the capital stock with 1993 as the base year. With reference to the
calculation method of Young18 and Zhang Jun et al., we take ten times the provincial industry-
specific fixed capital in 1993 as the provinces’ initial capital stock.
(b) Construction of the price indices of fixed assets investment. The price indices of fixed
assets investment are indicated by the price indices of provincial fixed assets investment. The
pre-1966 data are obtained by calculating provincial implicit investment deflators; for the
data after 1996, the price indices of fixed assets investment published by China Statistical
Yearbook are directly adopted. Finally, all the data are adjusted in a unified way with 1993 as
the base year.
(c) The depreciation rate of provincial industry-specific gross fixed capital formation is
fixed at 9.6 percent.
(d) Total social fixed asset investment is used for the current investment indicator.19 Data on
total social fixed asset investment are from the China Statistical Yearbook for various years
and the Statistical Yearbook of Chinese Investment in Fixed Assets 1950-1995 and other years.
In terms of management channels, investment in fixed assets may be divided into investment
in capital construction, investment in innovation and transformation, investment in real estate
development and investment in other fixed assets. This study classifies the investment in fixed
assets into three parts according to type of industries: investment from the private sector,
investment from the transport department of the public sector and investment from other parts
of the public sector.20 The data concerning all sorts of investment in fixed assets classified

17 Zhang Jun, Wu Guiying and Zhang Jipeng, “Estimation of China’s Provincial Material Capital
Stock: 1952-2000.” Zhang Jun et al. calculated data on China’s provincial material capital stock from
1952 to 2000 and we updated the database to 2009. See notes on web page http://www.cces.fudan.edu.
cn/ArticleDetail.aspx?ID=1174.
18 Alwyn Young, “Gold into Base Metals: Productivity Growth in the People’s Republic of China
during the Reform Period,” pp. 1220-1261.
19 When calculating China’s provincial material capital stock, Zhang Jun et al. used data on gross
fixed capital formation as the current investment indicator. These data are readily available in the China
Statistical Yearbook; but data on the industry-specific gross fixed capital formation of the provinces are
not available in any of the statistical yearbooks, so this study uses the available total investment in fixed
assets as an approximate value to substitute for gross fixed capital formation.
20 For the two periods 1993-2003 and 2004-2009, the China Statistical Yearbook has different
statistical items for industry-specific investment, so the contents of investment from the private
sector, the transport department of the public sector and other parts of the public sector differ for the
two periods. In the 1993-2003 period, the industries the private sector invested in include farming,
forestry, animal husbandry and fishery; mining; manufacturing; construction; wholesale and retail
trade; hotel and catering services; finance and insurance; real estate and other industries. Transport
infrastructure is not listed as a separate industry, being replaced by transport, warehousing, and posts and
telecommunications. The industries the other parts of the public sector invested in include the production
and supply of electricity, coal, gas and water; geological prospecting; water conservancy management;
34 Social Sciences in China

according to industry type are not available directly from the statistical yearbooks and must
be calculated and sorted separately. Specifically, the China Statistical Yearbook from 2005
to 2010 provides data on regional investment in fixed assets by industry classification which
can be directly sorted into data on investment in fixed assets by different industries. Prior to
2005, the China Statistical Yearbook did not directly provide data on regional investment
in fixed assets by industry classification, but it did provide data on regional investment in
capital construction and in innovation and transformation by industry classification. In the
sorting process, the author found that for all years, especially before 2000, judged by the data
on investment in capital construction from the three sources (the private sector, the transport
department of the public sector and other parts of the public sector), in most areas, especially
in the central and western regions, public sector investment in capital construction was greater
than that of the private sector, which means that investment in capital construction was
provided mainly by the government. If we substitute data on capital construction investment
for that on fixed assets investment we will not only underestimate the amount of capital input,
but also greatly underestimate investment from the private sector, neglecting the market’s
basic role in guiding investment in fixed assets. In addition, according to our ordering of
the data on investment in innovation and transformation from the three sources, the amount
of private sector investment is far higher than that of the public sector, which means that
investment in innovation and transformation comes mainly from private enterprises and that
the market plays a significant role. Accordingly, based respectively on investment in capital
construction and investment in innovation and transformation, we calculated investment from
the private sector and from the public sector transport department and other parts of the public
sector and then added up these two sets of data. In order to make the data on investment in
various industries closer to the data on investment in fixed assets, we added investment in
real estate development to the data on investment from the private sector. Due to limited data
resources, we did not take account of investment in other fixed assets in summing the data.
From the above method, we obtained data on the material capital stock of different
industries in various provinces in the period from 1993 to 2009. Accordingly, “transport
infrastructure” in this paper corresponds to the “material capital stock of the transport sector”
in the empirical analytical model.

social services, health, sports and social welfare; and state organs, party and government organs and
social organizations. For the 2004-2009 period, the industries the private sector invested in include
farming, forestry, animal husbandry and fishery; mining; manufacturing; construction; wholesale and
retail trade; hotel and catering services; finance; real estate; leasing and business services. Transport
infrastructure is made up of two parts: transport, warehousing and postal services; and information
transmission, computer services and software. The industries the other parts of the public sector invested
in include the production and supply of electricity, gas and water; scientific research, technical service
and geological prospecting; management of water conservancy, the environment and public facilities;
community and other services; education; health; social security and social welfare; culture, sports and
recreation; public administration and social organizations; and international organizations.
Zhang Xueliang 35

(2) The human capital variable (H ). This is a new economic growth factor, referring
mainly to expenditure on education, health, training, migration and information acquisition.
There are many ways to measure human capital stock; we use average years of schooling as
a proxy variable of human capital stock. The essence of this approach is to take “educational
attainment” as an indicator of the level of human capital (Barro and Lee21). Domestic scholars,
like Xu Xianxiang22 and Hao Rui,23 also use a similar method. In this article, the average
years of schooling H = 6s1+ 9s2+ 12s3+ 16s4, in which s1, s2, s3 and s4 represent respectively the
proportion of people with primary, junior secondary, senior secondary and tertiary education
in the population aged six years old and above. The data are from the Educational Statistics
Yearbook of China and China Statistical Yearbook published since 1993.
(3) Total exports (unit: 10,000 US dollars) export. New economic growth theory
internalizes export as a variable in the model. As a result of the learning-by-exporting (LBE)
effect, export enterprises are generally more productive than non-export enterprises.
(4) Regional policy policy. New economic growth theory holds that some of the economic
policies implemented by a government have an important impact on a country’s economic
growth (Frankel and Romer24). For a transitional economy like China, different policies will
tend to be an important factor for regional differences in a long time to come. Kanbur and
Zhang25 reviewed the impact of policy factors on regional economic development over the
past 50 years; Démurger et al.26 also emphasized the role of economic policy in regional
economic development in China since reform and opening up. This article sets up a dummy
variable, policy, to reflect changes in economic policy in China. Its value is set to 1 after 1999
when the grand strategy for the development of China’s west was put into effect and to 0
before 1999.
(5) Highway mileage (km) road. The cost of transportation is one of the three cornerstones
of the new economic geography. Traditional economics always assumes that space is
homogeneous and that the flow of commodities and services in space can be instantaneously
completed, while the new economic geography treats transportation costs as an endogenous
variable in economic growth. The cost of transportation is an important factor in decisions on
the siting of new plants, and highway mileage can be expected to have a positive impact on
economic growth.
(6) Level of urbanization urban. Urbanization and industrialization have been two major

21 R.J. Barro, and J.W. Lee, “International Data on Educational Attainment: Updates and
Implications,” pp. 541-563.
22 Xu Xianxiang and Shu Yuan, “Provincial Economic Growth Dynamics in China: 1978-1998.”
23 Hao Rui, “Economic Efficiency and Regional Equality: An Empirical Analysis of China’s Inter-
provincial Economic Growth and Inter-provincial Gaps (1978-2003).”
24 Jeffrey Frankel and David Romer, “Does Trade Cause Growth?”, pp. 379-399.
25 Ravi Kanbur and Xiaobo Zhang, “Fifty Years of Regional Inequality in China: A Journey through
Central Planning, Reform and Openness,” pp. 87-106.
26 S. Démurger et al., “Geography, Economic Policy and Regional Development in China,” pp. 146-
197.
36 Social Sciences in China

engines of economic growth in China since reform and opening up. With the advent of the
new century, urbanization entered a relatively independent stage of large-scale development.
The land factor was revalued and government “land-based finance” expanded public
infrastructure and promoted regional economic growth.27 Urbanization plays an important role
in regional economic growth. In this paper, the proportion of the non-agricultural population
in the total population is taken as an indicator of the level of urbanization (urban).
(7) Industrial agglomeration. Industrial agglomeration is one of the theoretical foundations
of the new economic geography, which holds that due to increasing returns to scale, even
if two areas have similar natural conditions, industrial agglomeration may take place in
one rather than the other owing to chance factors. Industrial agglomeration promotes
productivity and economic growth through the three micro-mechanisms of sharing, matching
and learning.28 In this article, LE (localization economies) and PE (Porter externalities) are
used to represent industrial agglomeration. “Localization economies” relates to the idea that
in a given area, enterprises can benefit from the economic activities of other enterprises in
the same industry, thus giving rise to a concentration of firms from the same industry in the
one place and promoting the growth of the industry in this area. Porter externalities, on the
other hand, relate to the idea that enterprises benefit chiefly from diversity, that knowledge
spillovers come mainly from enterprises in different industries rather than from enterprises
in the same industry, and that competition helps knowledge innovation and spillovers. Their
respective functional relations are expressed as follows:
g i / Yi
LE = (8)
∑ g / ∑Y
i
i
i
i
Ni / gi
PE = (9)
∑ Ni / ∑ gi
i i

in which g i represents the total industrial added value of province i, N i the number of
enterprises in the province i and Yi the GDP of the province i. The greater the value of the LE
index, the higher the degree of specialization of an industry, and the more conducive it is to
knowledge spillovers and regional economic growth. The value of PE is negatively correlated
with the degree of market competition in an industry: the lower the value, the stronger the
competition.

IV. Model Estimation Results

1. Model specification and estimation


Equation (10) takes into consideration not only the factors of neoclassical economic

27 Research Group on China’s Economic Growth (CASS), “Urbanization, Fiscal Expansion and
Economic Growth.”
28 G. Duranton and D. Puga, “Micro-foundations of Urban Agglomeration Economics.”
Zhang Xueliang 37

growth, new economic growth and new economic geography, but also the spatial spillover
effects of economic growth itself and transport infrastructure, and brings the capital stock of
transport infrastructure in other regions into the model:
ln(Yit)=α0+ρWln(Yit)+α1ln(Lit)+α2ln(Kcit)+α3ln(Ktit)+α4ln(Kgit)+α5ln(Hit)
+α6ln(exportit)+α7policyit+α8ln(roadit)+α9ln(urbanit)+α10ln(LEit)
+α11ln(PEit)+α12ln(OKtit)+μi+αt+εit (10)
The composite error term in Equation (10) is μi+αt+εit, a dual-factor model, that takes
account of two effects: the individual effect μi and the time effect αt. Based on the four types
of spatial weight matrix constructed above, this part uses the fixed effects spatial lag model
to estimate separately the expanded model (10) that includes the four types of matrix. The
estimation results are shown in Table 1.

Table 1 Estimation Results of the Spatial Weight Matrix Model within the Research
Framework of Multi-dimensional Economic Growth (1993-2009)
Model omitting Transport Population
Estimation Binary Per capita GDP
spatial spillover network density
method Wcont (2) Wpergdp (5)
effects (1) Wnet (3) Wperpop (4)
— 0.124** 0.113** 0.131** 0.143**
W ln Y
— 2.24 1.99 2.05 2.12
0.257*** 0.210*** 0.214*** 0.215*** 0.214***
lnL
5.24 6.32 6.54 7.22 6.64
0.131*** 0.112** 0.118* 0.115** 0.115**
lnKc
4.25 2.21 2.24 2.21 1.99
0.058*** 0.030*** 0.028** 0.029** 0.031**
lnKt
6.25 2.57 2.43 2.41 2.38
0.098*** 0.089*** 0.090*** 0.092** 0.095**
lnKg
4.14 2.95 3.38 2.16 2.32
0.074** 0.065*** 0.067** 0.070** 0.064**
lnH
2.20 2.89 1.99 2.23 2.53
0.082*** 0.071* 0.076 0.075* 0.075
lnexport
3.23 1.84 1.59 1.71 1.50
0.083*** 0.142*** 0.143*** 0.131*** 0.143***
policy
7.33 6.75 6.53 6.32 6.14
0.052*** 0.023** 0.020*** 0.023*** 0.022***
lnroad
2.72 2.53 2.63 2.72 2.70
0.103** 0.092** 0.095 0.093** 0.091*
lnurban
2.48 2.22 1.52 2.24 1.87
38 Social Sciences in China

0.045** 0.051 0.052 0.053* 0.050*


lnLE
1.96 1.57 1.53 1.64 1.63
-0.053** -0.032** -0.031** -0.032** -0.034**
lnPE
-2.32 -2.24 -2.19 -2.21 -2.20
— 0.031*** 0.010** -0.007* 0.005**
lnOKt
— 3.43 2.45 -1.63 2.54
Constant -3.532*** -2.543*** -2.565*** -2.563*** -2.553***
term -6.53 -7.45 -7.34 -6.73 -6.97
log L 165.25 164.45 174.54 165.35 170.39
No. of
493 493 493 493 493
samples
Notes: 1. Spatial samples are from 29 provinces, municipalities and autonomous regions on the
mainland of China. Chongqing’s data are incorporated into Sichuan province. Data for the Tibet
Autonomous Region are incomplete and not included in the samples.
2. Figures under parameter estimations are z test values; *, ** and *** represent levels of
significance at 10%, 5% and 1% respectively.

It should be noted that if we were to use the ordinary least squares (OLS) method to
estimate Equation (10), the estimated value of the coefficient would be biased or invalid,
so we use the maximum likelihood (ML) method for the estimation. The results are shown
above.
In the empirical analysis we also take into account the fact that since 1999, China has
adopted grand strategies for the development of the western region, the revitalization of old
industrial bases in the eastern region, and the rise of the central region, and that accession to
the WTO in 2001 further broadened China’s opening to the rest of the world. Moreover, the
government has increased investment in transport infrastructure over recent years. All these
developments have significantly changed the economic environment of this country. Against
this backdrop, we selected data for 2000-2009 and, after eliminating the economic policy
variable (policy), re-estimated Equation (10). The estimation results are shown in Table 2.

Table 2 Estimation Results of the Spatial Weight Matrix Model within the Research
Framework of Multi-dimensional Economic Growth (2000-2009)
Model omitting Binary Transport Population Per capita GDP
Estimation
spatial spillover Wcont (7) network density Wpergdp (10)
method
effects (6) Wnet (8) Wperpop (9)
— 0.157*** 0.140** 0.153** 0.161**
W ln Y
— 2.87 2.53 2.46 2.25
Zhang Xueliang 39

0.292*** 0.284*** 0.282*** 0.294*** 0.285***


lnL
8.33 10.25 8.95 8.29 8.37
0.125*** 0.105*** 0.098*** 0.102*** 0.105**
lnKc
4.81 3.32 3.34 3.28 2.28
0.048*** 0.025*** 0.022*** 0.024** 0.025**
lnKt
4.26 3.35 2.85 2.46 2.28
0.127*** 0.113*** 0.112*** 0.118** 0.120***
lnKg
3.2 2.83 3.48 2.23 2.89
0.102*** 0.092** 0.094** 0.098** 0.095**
lnH
2.68 2.21 1.96 2.21 2.35
0.120*** 0.101* 0.092* 0.109* 0.115**
lnexport
3.22 1.68 1.65 1.87 2.36
0.084** 0.043** 0.047** 0.045** 0.050**
lnroad
2.50 2.24 2.35 2.44 2.41
0.085* 0.070** 0.081** 0.077* 0.073*
lnurban
1.78 2.54 1.98 1.68 1.69
0.062* 0.071 0.072* 0.081* 0.078*
lnLE
1.76 1.59 1.93 1.94 1.93
-0.034** -0.027** -0.021** -0.019** -0.018**
lnPE
-2.34 -2.24 -2.35 -2.37 -2.36
— 0.042*** 0.028*** -0.01*** 0.022***
lnOKt
— 3.29 2.88 -3.65 2.87
Constant -5.265*** -4.553*** -4.578*** -4.432*** -4.523***
term -10.56 -8.58 -7.54 -7.43 -7.75

log L 245.26 250.34 260.56 255.30 258.92

No. of
290 290 290 290 290
samples
Notes: Same as Table 1.

2. Estimation results and interpretation


(1) Analysis of the role of transport infrastructure in regional economic growth is an
important element in this study. We chose two variables to reflect the impact of the local
transport infrastructure on local economic growth, i.e., the transport infrastructure variable
Kt and the highway mileage variable road. In the new economic geography, the latter factor is
related to transport infrastructure; it reflects transport capacity. Our empirical findings show,
first, the coefficients of these two variables are both positive. For example, in the per capita
40 Social Sciences in China

GDP spatial weight matrix model in Table 1, the results of our empirical analysis show that
the output elasticity of the two variables that reflect transport capacity is 0.031 and 0.022
respectively. This indicates that the improvement of transport capacity enhances the ability of
the local area to communicate with the outside world and reduces transportation costs, thus
contributing to regional economic growth. Since the beginning of the new century, as shown
in Table 2, the sum of the output elasticity of the two variables reflecting transport capacity
has increased to about 0.07 and the role of transport infrastructure in economic growth has
been more striking. Secondly, in both Table 1 and Table 2, compared with the original models
(1) and (6), which do not take into account spatial spillover effects, the estimation results of
the four types of spatial weight matrix model show that the values of the output elasticity of
transport infrastructure Kt are all lower by about 0.03 than those calculated on the basis of
the original model, which did not consider the spatial spillover effects. This suggests that if
a model fails to consider the spatial spillover effects, the role of transport infrastructure in
regional economic growth will be overestimated.
(2) Our focus in this article is to ascertain the role of the transport infrastructure of other
regions in the economic growth of a particular region, i.e., the spillover effects of transport
infrastructure. Both Table 1 and Table 2 list the estimated coefficients of the four models
created respectively on the basis of the binary spatial weight matrix Wcont, the population
density spatial weight matrix Wperpop, the per capita GDP spatial weight matrix Wpergdp and
the transport network spatial weight matrix Wnet. For example, in the samples as a whole for
the period 1993-2009, in the models based on the binary spatial weight matrix, the transport
network spatial weight matrix and the per capita GDP spatial weight matrix, the output
elasticity of transport infrastructure for other regions is respectively 0.031, 0.010 and 0.005.
In the early years of the new century, the three types of output elasticity rose respectively
to 0.042, 0.028 and 0.022, that is, transport infrastructure in other regions had a positive
spatial spillover effect on local economic growth. Transport infrastructure integrates regional
economic activity throughout China, and through the diffusion effect, the development of one
region provides impetus for the development of adjacent regions. Consequently, the transport
infrastructure of other regions produces a positive spatial spillover effect on the economic
growth of a particular region.
It needs to be pointed out that in the empirical model created using the per capita GDP
spatial weight matrix W pergdp, transport infrastructure in other regions does not exhibit
negative spatial spillover effects on local economic growth as might have been expected
from the theoretical model. This finding may be related to the existence of significant spatial
agglomeration in the course of China’s regional economic growth. In this country, provinces
with similar levels of GDP compete with each other to attract the factors of production, but
the spatial agglomeration effect may be more significant. Regional economic theory tells
us that economic activities exhibit spatial dependence in geographical space, i.e., economic
growth in one area may have an impact on economic growth in neighboring areas, which may
Zhang Xueliang 41

then have demonstration, promotion and imitation effects on one another. For example, in
the Yangtze Delta region, Shanghai, Jiangsu and Zhejiang have maintained a similarly rapid
pace of growth. As affluent areas are surrounded by other affluent areas, they tend to cluster
together in space,29 and the development of inter-regional transport infrastructure further
strengthens this spatial dependence. As a result, in the empirical model based on the per capita
GDP spatial weight matrix Wpergdp, transport infrastructure in other regions produces positive
spatial spillovers on economic growth in a particular region.
The spatial spillover effects of regional economic growth also corroborate the above view.
In Table 1 and Table 2, all the values of the coefficient ρ reflecting the spatial spillover effects
of regional economic growth are significantly positive, indicating that Chinese provincial
economic growth shows significant spatial dependence: a region’s economic growth is
influenced by the economic growth of adjacent regions and, for a particular province, faster
development in neighboring provinces augurs well for the province itself.
In addition, we find the evidence of the negative spillover effects of transport infrastructure
in the model based on the population density spatial weight matrix Wperpop. With this model,
in Table 1, the output elasticity of transport infrastructure in other regions is -0.007, which
means that for each one yuan increase in the transport infrastructure capital stock of the other
regions, local economic growth falls by 0.007 yuan. Since the advent of the new century,
this evidence has become clearer. In Table 2, the output elasticity of transport infrastructure
in other regions is -0.010 and is statistically more significant. We believe this conclusion is
in line with the above theoretical expectations and conforms to the reality of China’s one-
way regional migration flows. The household registration system restricts the population
flow, but great progress has been made in the development of a unified labor market in China
since the 1990s and there is now much more freedom of labor mobility. Due to transport
infrastructure’s network properties and spatial externalities, the development of inter-regional
transport infrastructure may accelerate the cross-regional flow of labor. The accelerated
development of inter-regional transport infrastructure will enhance the siphoning effect
whereby developed regions suck the factors of production from backward regions. This is
especially so when the investment environment, urbanization and human capital qualities of
the backward regions are yet to be upgraded. At present, China’s cross-regional labor mobility
as a whole has the character of a one-way movement from the backward western and central
regions to the developed coastal areas in the east. For example, according to data from the
Fifth National Population Census in 2000, 82.1 percent of the total migrant population moved
from the central and western regions toward the east and only 17.9 percent flowed within the
confines of the eastern region. The major source provinces for migration were Jiangxi, Anhui,
Hubei, Hunan and Henan in the central region, and Sichuan, Chongqing, and Guangxi in the
western region while Guangdong, Shanghai, Zhejiang, Fujian, Jiangsu and Beijing were net

29 Zhang Xueliang, “A Spatial Econometric Analysis of Regional Economic Convergence in China.”


42 Social Sciences in China

recipients of migrant workers. Development of inter-regional transport infrastructure reduced


the transportation cost of labor mobility, resulting in a faster movement of the various factors
of production, including labor and human capital, from the backward areas to the developed
areas, thus further weakening local economic development.30 This situation has not changed
radically in recent years.
(3) Among all other factors, labor and other public sector capital still make a very great
contribution to regional economic growth. Although the role of the labor input in economic
growth has weakened somewhat since 2000, it is still that factor that makes the greatest
contribution to economic growth. Moreover, there is still room to further enhance the role of
the capital stock of other parts of the public sector in economic growth. This means that the
role in economic growth of China’s uniquely advantageous labor resources and government
departments’ capital stock should continue to receive our attention. Among the new economic
growth factors, human capital has made significant contributions to regional economic growth,
especially since the advent of the new century. Exports have also played a positive role in
promoting economic growth. It should be stressed that regional policy plays a significant
role in economic growth, indicating that in the period of transition to a market economy in
China, the central government’s regional policy and strategy have a strong effect on economic
development. Especially since reform and opening up, against the backdrop of a widening gap
between regions, the central government has issued a series of policies favoring the central and
western regions in order to reduce regional disparities. The public capital stock of the backward
central and western regions makes up a high and stable proportion of their total capital stock.
Among the new economic geography factors, localized economies foster regional economic
growth, showing that knowledge accumulation arising out of continuous exchanges among
firms in the same industry promotes regional concentration of production and strengthens
interaction among individuals and firms with common goals and similar abilities, thereby
enhancing regional productivity and promoting economic growth. The coefficient of Porter
externalities is negative, indicating that the diversification of local industries can also promote
regional economic growth. The more competitive the local industrial structure, the more

30 The anonymous peer reviewers have pointed out that the impact of transport infrastructure on
different spatial units may be different in degree and magnitude and may be either positive or negative,
and that the estimated coefficients are bound to be different. They suggested that this article study the
concrete spatial spillover effects of transport infrastructure on some specific regions on the basis of the
overall analysis of regional spillover effects in general. This valuable suggestion that points the way
for advanced spatial econometric research. In fact, analysis of the correlations between the two factors
in a specific spatial unit and calculation of their coefficients is quite a cutting-edge research project in
the field of spatial econometrics. From a methodological point of view, we can try to create a spatial
econometric model to study the interactions of the two factors in a specific spatial unit, but the biggest
difficulty encountered in the estimation of a specific model is that it requires estimation of too many
parameters, so many that there may be more parameters than samples. To solve this problem we need to
find a good method of data dimension reduction; this represents an important piece of follow-up research
for our study.
Zhang Xueliang 43

conducive it is to the development of production. In addition, it needs to be particularly


emphasized that urbanization is still one of the main engines driving China’s economic
growth; as shown in Table 2, the output elasticity of Chinese urbanization is about 0.08. On
the one hand, Chinese urbanization is now accelerating. This is especially true of the first
years of the new century; calculated on the basis of the proportion of the urban population
in the total population, China’s population urbanization has averaged a 1.37 percentage
point increase per annum. The rate of land urbanization has been even faster: the built-up
area increased from 24,000 km2 in 2001 to 38,000 km2 in 2009, an increase of nearly 60
percent.31 During this process, the land factor was revalued, and the local governments used
revenues from “land-based finance” to increase their investment in capital-intensive public
infrastructure directly related to land urbanization. As Table 2 shows, the contribution of
public infrastructure to regional economic growth was two percentage points higher for this
period than for the sample period as a whole. On the other hand, the inter-regional flow of
factors of production mostly proceeds between cities and important transport infrastructure
also mainly links cities. Consequently, the accelerated urbanization process has objectively
augmented the spatial spillover effects of transport infrastructure. It can also be observed
from Table 2 that in the new century, in the empirical models based respectively on the binary
spatial weight matrix, the transport network spatial weight matrix and the per capita GDP
spatial weight matrix, the output elasticity of the transport infrastructure in other regions has
been greater than that for the sample period as a whole.

V. Conclusions and Policy Implications

A model of the spatial synergistic effects of multidimensional factors will overestimate the
role of transport infrastructure in regional economic growth if it does not take into account
the spatial spillover effects. This shows that the introduction of spatial dimensional analysis
is of great significance for a dialectical understanding of the relationship between transport
infrastructure and economic growth. In the 12th Five-year Plan and for an even longer period
in the future, China will continue to increase investment in infrastructure. In the macro context
of accelerated regional economic integration and coordinated urbanization, we should not
only pay attention to the total growth effects of government investment, including investment
in transport infrastructure, but also, and more importantly, should attach importance to
the spatial efficiency of the transport infrastructure capital stock, thereby enhancing the
effectiveness of regional transport policy.
Once we allow for the spatial spillover effects, the contribution of transport infrastructure

31 China’s population urbanization rate is calculated on the basis of the data from the China Statistical
Yearbook of 2011. Over the same period, the urban population increased from 480 million to 670
million, a rise of only 34%. The data on land urbanization are from the Research Group on China’s
Economic Growth (CASS), “Urbanization, Fiscal Expansion and Economic Growth.”
44 Social Sciences in China

to local economic growth falls, but, taken in conjunction with the output elasticity of the
highway mileage variable, which is related to transport infrastructure and reflects transport
capacity, the output elasticity of transport infrastructure still remains at 0.05 or so. Especially
in the 21st century, the output elasticity of transport infrastructure in regional economic
growth has risen to about 0.07, which indicates that at this stage transport infrastructure
capital stock still plays an important role in promoting economic growth in China. Therefore,
in order to cope with the impact of the international financial crisis on economic growth and
maintain the stable and rapid development of the domestic economy, the Chinese government
must continue to use transport and other infrastructure investment as an important means of
implementing a proactive fiscal policy.
With regard to inter-regional economic relations in China, the spatial spillover effects
of transport infrastructure in other regions on economic growth in a particular region are
predominantly positive. The development of transport infrastructure as a link for regional
economic and social activities promotes the cross-regional flow of the various factors of
production and regional economic integration. Moreover, regional economic growth manifests
clear spatial dependency and the spatial spillover effects of economic growth accelerate the
development of transport infrastructure, helping economic growth in both the local area and
the other areas. But we also found evidence in the model based on the population density
spatial weight matrix of negative spatial spillovers related to the high mobility and one-
way movement of population. The development of transport infrastructure accelerates the
flow of population to the economically developed areas, thus inhibiting economic growth
in less developed areas. This finding has strong policy implications: while devoting major
efforts to the development of transport facilities and other infrastructure, local governments
must also endeavor to improve the local investment environment, increase the local level of
urbanization, upgrade human capital and provide more and better employment opportunities;
only in this way will it be possible to reduce the outflow of labor and human capital and
achieve faster local economic growth from the development of transport infrastructure.
Meanwhile, since the inter-regional flow of factors of production is mainly between the
cities and important transport infrastructure mainly links (large) cities with one another, we
should continue to proceed with the urbanization drive in order to expand the positive spatial
spillover effects of transport infrastructure.
Of course, while developing inter-regional transport infrastructure to a high standard, China
must also accelerate the construction of transport infrastructure within the eastern, the central
and the western regions, especially in the city clusters of these regions. The eastern region
should endeavor to achieve the qualitative upgrading of transport infrastructure on the basis of
the expansion of transport stock and accelerate the process of regional transport integration;
the central and western regions still need, in view of the small scale of their current transport
infrastructure, to focus on quantity and speed up internal transport infrastructure construction
as well as actively promoting the intra-regional flow of various factors of production,
Zhang Xueliang 45

especially intra-regional labor flows and employment, in order to fully exploit China’s labor
resource advantage. With regard to the policy of regional development, we believe that
even within the central and western regions, we should also follow a policy of coordinated
regional development while pursuing agglomeration. The empirical results of this study
show that industrial agglomeration significantly promotes regional economic growth and that
both specialization and diversification of local industries have a positive effect on regional
economic growth. The central and western regions should therefore strive to foster new
regional growth poles, quicken the pace of industrial agglomeration in geographical space and
promote the development of a city cluster economy within their spheres of administration.
This will also create favorable conditions for bringing into full play the positive spatial
spillover effects of transport infrastructure on regional economic growth within the city
clusters in the central and western regions.
Finally, it needs to be pointed out that this study has calculated the coefficient of the “capital
stock of transport infrastructure in other relevant regions OKt,” proving beyond any doubt
the existence of global spatial correlation; in other words, we have overall found evidence of
significant spatial spillover effects of transport infrastructure on regional economic growth.
However, we still need to undertake further analyses of specific regions and find how much
and in what way they are affected by transport infrastructure—this is a more complex and
highly significant research topic.

Notes on Contributor

Zhang Xueliang, Ph.D., is Associate Research Fellow of the Institute of Finance and Economics of
Shanghai University of Finance and Economics and a Shuguang Scholar of Shanghai. His academic
interests are urban economics, regional economics, spatial statistics and spatial econometrics. Over
recent years, he has presided over many important research projects. His representative works include:
“Economic Convergence and Its Action Mechanism in the Yangtze River Delta: 1993-2006” (长三
角地区经济收敛及其作用机制:1993-2006, The World Economy [世界经济], 2010, no. 3) and
“The True Source of Economic Growth in the Yangtze River Delta: Capital Accumulation, Improved
Efficiency or TFP Contribution?” (探寻长三角地区经济增长的真正源泉:资本积累、效率改善抑或
TFP贡献?, China Industrial Economics [中国工业经济], 2009, no. 5, co-authored with Sun Haiming).
E-mail: itiszxl@yahoo.com.cn.

References

Aschauer, David Alan. “Is Public Expenditure Productive?”. Journal of Monetary Economics, vol. 23,
March 1989, no. 2.
Banister, D. and Y. Berechman. “Transport Investment and the Promotion of Economic Growth.”
Journal of Transport Geography, vol. 9, 2001, no. 3.
Barro, R.J. and J.W. Lee. “International Data on Educational Attainment: Updates and Implications.”
46 Social Sciences in China

Oxford Economic Papers, vol. 53, July 2001, no. 3.


Bavaud, Francois. “Models for Spatial Weights: A Systematic Look.” Geographical Analysis, vol. 30,
January 1998, no. 1.
Boarnet, M.G. “Spillover and the Locational Effects of Public Infrastructure.” Journal of Regional
Science, vol. 38, March 1998, no. 3.
Bonaglia, F., E. La Ferrara and M. Marcellino. “Public Capital and Economic Performance: Evidence
from Italy.” IGIER Working Paper no. 163, Feb. 2000.
Cantos, Pedro, Mercedes Gumbau-Albert and Joaquín Maudos. “Transport Infrastructures, Spillover
Effects and Regional Growth: Evidence of the Spanish Case.” Transport Reviews, vol. 25, January
2005, no. 1.
Démurger, S. et al. “Geography, Economic Policy and Regional Development in China.” Asian
Economic Papers, vol. 1, Winter, 2002, no. 1.
Duranton, G. and D. Puga. “Micro-foundations of Urban Agglomeration Economics.” In Handbook of
Regional and Urban Economics, vol. 4, eds. J.V. Henderson and J.F. Thisse. Amsterdam: Elsevier-
North Holland, 2004.
Frankel, Jeffrey and David Romer. “Does Trade Cause Growth?”. American Economic Review, vol. 89,
June 1999, no. 3.
Fujita, M., P. Krugman and A.J. Venables. The Spatial Economy: Cities, Regions, and International
Trade. Cambridge, MA: MIT Press, 1999.
Gao, Feng. Transport Infrastructure Investment and Economic Growth (交通基础设施投资与经济增
长). Beijing: China Financial & Economic Publishing House, 2005. 
Hao, Rui. “Economic Efficiency and Regional Equality: An Empirical Analysis of China’s Inter-
provincial Economic Growth and Inter-provincial Gaps (1978-2003)” (经济效率与地区平等:中国
省际经济增长与差距的实证分析 [1978-2003]). World Economic Papers (世界经济文汇), 2006,
no. 2.
Kanbur, Ravi and Xiaobo Zhang. “Fifty Years of Regional Inequality in China: A Journey through
Central Planning, Reform and Openness.” Review of Development Economics, vol. 9, Feb. 2005, no.
1.
Nurkse, Ragnar. Problems of Capital Formation in Underdeveloped Countries. Oxford: Basil Blackwell,
1953.
Research Group on China’s Economic Growth (CASS). “Urbanization, Fiscal Expansion and Economic
Growth” (城市化、财政扩张与经济增长). Economic Research Journal (经济研究), 2011, no. 11.
Rosenstein-Rodan, Paul. “Problems of Industrialization of Eastern and Southeastern Europe.” The
Economic Journal, vol. 53, Jun.-Sep. 1943, no. 210/211.
Rostow, W.W. The Stages of Economic Growth: A Non-Communist Manifesto. Cambridge: Cambridge
University Press, 1960.
Wong, David W.S. and Jay Lee. Statistical Analysis of Geographic Information with ArcView GIS and
ArcGIS. Trans. Zhang Xueliang. Beijing: China Financial & Economic Publishing House, 2008.
Xu, Xianxiang and Shu Yuan. “Provincial Economic Growth Dynamics in China: 1978-1998” (中国省
Zhang Xueliang 47

区经济增长分布的演进 [1978-1998]). China Economic Quarterly (经济学 [季刊]), 2004, no. 3.


Young, Alwyn. “Gold into Base Metals: Productivity Growth in the People’s Republic of China during
the Reform Period.” Journal of Political Economy, vol. 111, Dec. 2003, no. 6.
Yu, Yongjun and Lu Yuqi. “A Study of the Relationship between Transport Investment and Economic
Development and Methods of Evaluating Its Regional Effects” (交通投资与经济发展的关系及其区
域效应评价方法研究). Human Geography (人文地理), 2005, no. 1.
Zhang, Jun, Wu Guiying and Zhang Jipeng. “Estimation of China’s Provincial Material Capital Stock:
1952-2000” (中国省际物质资本存量估算:1952-2000). Economic Research Journal, 2004, no. 10.
Zhang, Xueliang. “A Regional Comparative Analysis of Transport Infrastructure and Economic Growth
in China” (中国交通基础设施与经济增长的区域比较分析). Journal of Finance and Economics (财
经研究), 2007, no. 8.
——. “A Spatial Econometric Analysis of Regional Economic Convergence in China” (中国区域经济
收敛的空间计量分析). Journal of Finance and Economics, 2009, no. 7.

—Translated by Lin Hong from


Zhongguo Shehui Kexue (中国社会科学), 2012, no. 3
Revised by Sally Borthwick

You might also like