Transport Infrastructure, Urban
Growth and Market Access
in China
Nathaniel Baum-Snow, Brown University
Loren Brandt, University of Toronto
Vernon Henderson, LSE
Matthew Turner, Brown University
Qinghua Zhang, Peking University
Questions of Interest
• How have new highways and railroads influenced urban growth in
Chinese prefectures and cities?
• Mechanisms internal to cities
• Improved access to domestic markets
• Improved access to international markets
• To what extent are these influences reflected in growth versus
redistribution?
Motivations
• Important Policy Questions
• Developing countries are making huge investments in highways and
railroads - about $200 billion per year in China since 1990
• The impacts of new transport infrastructure in developing country
contexts are just starting to receive serious research attention
• About 20% of World Bank lending is for transport infrastructure
• Little is known about the relative importance of various classes of
mechanisms through which transport infrastructure may drive urban growth
• New highways typically influence internal city transport costs, those
between cities and international transport costs simultaneously
Potential Channels
• Mechanisms internal to cities
• Free up low agglomeration spillover land intensive activities to
decentralize
• Enhance agglomeration spillovers across firms conditional on
locations
• Lower cost access to shared intermediate inputs
• Easier flow of ideas across firms
• Allow cities to accommodate greater population, thereby
enhancing their productivity through agglomeration economies
• Improved access to domestic markets and international markets
• Ricardian gains from trade
• Knowledge spillovers
• Innovation through tougher competition
• Promote structural change
Preview of Results
• New highways promoted more GDP growth in prefectures that became better
connected to more populous and prosperous domestic regions
• We find less robust commensurate effects for population
• New highways promoted population growth in prefectures that became
better connected to international ports
• Because we find no commensurate effects for GDP, most of this looks like
redistribution rather than growth
• Still represents welfare gains for many people because of higher GDP
levels in regions with improved port access
• Radial railroads serving prefecture cities promoted population growth but not
statistically significant GDP growth
• Once again, looks like welfare enhancing population redistribution
Relationship to the Literature
• A burgeoning literature investigates effects of transportation on regional
growth in developing countries
• Alder (2014), Banerjee, Duflo & Qian (2014), Bosker (2014), Donaldson
(2014), Donaldson & Hornbeck (2014), Faber (2014), Fajgelbaum &
Redding (2014), Ghani, Goswami & Kerr (2014), Storeygard (2013)
• Another literature examines how within city transport infrastructure
improvements can promote urban growth
• Duranton & Turner (2012), Baum-Snow (2014), Baum-Snow et al. (2014)
• This is the first study to examine both classes of explanations simultaneously,
and with a new identification strategy
Context
• Our primary analysis starts in 1990:
• China is still a planned economy, except for a small amount of
within-prefecture free trade in agriculture
• Inter-prefecture trade proscribed by provincial authorities
• Almost all inter-prefecture shipments by rail, mostly between
prefecture cities and the provincial capitals
• Mostly unpaved roads used only for within-prefecture transport of
agricultural goods from farms and manufacturing goods to farms
• By 2010:
• Largely a free market economy, with land and trade flows allocated
by the market
• Largely complete national express highway system
Trends in Outcomes of Interest
1962/1982
1990
2010
Change
1990-2010
6.8
(0.9)
15.1
(0.7)
3.04
(0.43)
0.13
(0.20)
5.9
(1.2)
13.9
(0.8)
3.20
(0.56)
0.35
(0.27)
Prefecture Level Outcomes
log GDP
log Population
14.8
(0.7)
3.8
(0.8)
14.9
(0.6)
2010 Definition Prefecture City Outcomes
log GDP
log Population
13.4
(0.8)
2.7
(1.1)
13.5
(0.8)
• Very rapid GDP growth everywhere
• Despite relative per-capita GDP in rural versus urban areas of 2.4 in 1990 and 2.1 in
2010, only 38% of the population lives in prefecture cities
• Hukou migration restrictions are an important element of the context
1990 Road Network
1999 Road Network
2010 Road Network
1962 Road Network and 5-7 Plan
Expansions in Transport Infrastructure Serving Cities
Road Rays
Rail Rays
1962
1990
2010
Change
1990-2010
2.0
(1.4)
1.1
(1.2)
2.9
(1.4)
1.5
(1.3)
3.8
(2.0)
1.8
(1.3)
3.8
(2.0)
0.3
(1.1)
Measures of Explanatory Variables
• Intuitive variants of “Market Potential” based on Fujita, Krugman &
Venables (1999)
• GDP within 6 hours by road (excluding own prefecture)
• Employment within 6 hours by road (excluding own prefecture)
• Travel time to the nearest international port
• GDP and employment variants of “Market Access” based on Eaton &
Kortum (2002) and Hornbeck & Donaldson (2014)
• Smooth measure of connection to other markets that incorporates
recursion and transitivity of linkages between regions
• Domestic component
• International component
• Radial highways and railroads serving prefecture cities
Correlation Across 2010 MP and MA Measures
Measure Used for Analysis
Market Potential, GDP
Market Potential, Employment
GDP Market Access, q=1.5, r=0.2
Pop Market Access, q=1.5, r=0.2
Market Potential
Exp or
Emp
GDP
1.00
0.93
0.41
0.35
0.93
1.00
0.43
0.41
GDP Market Access
q=5
q = 1.5
q=5
r = 0.2 r = 0.02 r = 0.02
0.69
0.72
0.86
0.83
0.14
0.15
0.93
0.93
0.20
0.22
0.96
0.95
Pop Market Access
q=5
q = 1.5
q=5
r = 0.2 r = 0.02 r = 0.02
0.57
0.68
0.84
0.87
• All MA measures mutually highly correlated
• MA and MP measures are capturing something somewhat different
0.12
0.15
0.93
0.93
0.18
0.21
0.95
0.96
Population MP and MA
Market Potential
Market Access
Population MP and MA
Market Potential
Market Access
Population MP and MA
Market Potential
Market Access
Components of 2010 Market Access
Growth of MP and MA
1990
2010
Change
1990-2010
Market Potential Measures
log GDP Within 6 Hours
log Employment Within
6 Hours (millions)
log Road Time to Nearest
Port
4.7
(1.2)
15.65
(1.3)
7.16
(1.5)
9.9
(1.3)
17.29
(1.2)
5.9
(1.3)
5.3
(0.9)
2.01
(0.7)
-1.3
(0.4)
Market Access Measures
log Total Market Access
(GDP)
log Domestic Market
Access (GDP)
log External Market
Access (GDP)
log Total Market Access
(Population)
log Domestic Market
Access (Population)
log External Market
Access (Population)
4.78
(0.17)
4.59
(0.17)
3.01
(0.21)
7.66
(0.15)
7.66
(0.15)
-0.02
(0.21)
6.56
(0.16)
6.27
(0.16)
5.17
(0.18)
7.81
(0.14)
7.80
(0.14)
3.66
(0.18)
1.78
(0.06)
1.68
(0.06)
2.16
(0.09)
0.15
(0.06)
0.14
(0.06)
3.68
(0.09)
• Average prefecture GDP
growth of 3.0 log points
• Average prefecture
population growth of
0.13 log points
• Versions of MA and MP
built using 2010 roads
and 1990 quantities are
quite similar to 2010
numbers
• Therefore, most of the
MP and MA growth is
from improvements in
the road system
Empirical Strategy
• Think about a world with no trade and no highways in 1990, but some
unobservables about prefectures that influence growth but are fixed
over time
• This implies estimating coefficients in a regression equation like the
following:
 t yi = a0  a1 Rit  b1 Ait  b2 Eit  CX i  vit
population
or GDP
growth rate
radial highways access to
and railroads
domestic
serving
markets
prefecture city
access to
international
markets
predetermined
control
variables
• Still a potential problem of unobservables (not in X) correlated with R,
A and E that drive growth
Identification Challenges
• Highways built to serve growing cities
• Highways built to connect more productive or export oriented cities
• Unobservables that drive regional growth also predict modern highway
construction
• Cities with more railroads have more industry
Instruments
• 2010 radial highways and railroads serving prefecture cities
• 1962 radial roads and railroads serving prefecture cities
• Measures of market potential and market access along the 2010 road
and highway network
• 1962 roads and the 5-7 Plan interacted with 1982 expenditure or
1982 employment
• Road travel time to nearest port
• 1962 roads and 5-7 Plan travel time to nearest port
• International market access
• External market access along 1962 roads and the 5-7 Plan with
1990 exports
Instrument Validity
• 1962 roads
• Existence of these rights of way made subsequent highway construction
less costly
• Less than half of such roads were even passable by truck, and the ones
that were passable had top speeds of about 25 kph
• Mostly unpaved and existed almost exclusively for local transport of
agricultural goods
• Controls for variables correlated with the amount of prefecture
agricultural activity are therefore crucial
• 1962 railroads
• Existed for shipping manufactures and agricultural products mostly
between prefecture cities and provincial capitals
• 5-7 Plan
• Developed in the early 1980s to link 38 nodal prefecture cities in our study
area
• Idea is to look at cities in between which ultimately got linked into the
network only for this reason
Sources of Identifying Variation
• Compare prefectures receiving more limited access highways between
1990 and 2010 because of having had more unimproved roads serving
the city in 1962 relative to those with fewer unimproved roads in 1962
• Compare prefectures better linked to larger cities because they were
intermediate locations on the 5-7 Plan or central in the 1962 road
network to others
• Compare prefectures better linked to international ports because they
were intermediate locations on the 5-7 Plan or central in the 1962 road
network to others not linked well to ports on these networks
• All of these are comparisons conditional on controls for
• provincial capital indicators
• 1982 prefecture population
• 1982 prefecture industry mix
• 1982 prefecture education
• distance to the coast
Population MP and MA
Market Potential
Market Access
Instruments for Population MP and MA
uses 1962 road network+5-7 plan and 1982 employment or population
Market Potential
Market Access
First Stage Results – Highways, Railroads and MP
1962 Road Rays
1962 Rail Rays
log 1962/1982 Expend.
Market Potential
log 1962/1982 Emp.
Market Potential
log 1962 Time to
Nearest Port
Provincial
Capital
ln 1982 Prefec
Pop
1982 Fraction
HS Graduate
1982 Fraction
Manufacturing
log km to Coast
Constant
R-squared
Road Rays
0.35***
(0.07)
0.18*
(0.09)
-0.00**
(0.00)
Rail Rays
-0.01
(0.05)
0.33***
(0.07)
0.00
(0.00)
-0.24*
(0.12)
1.74***
(0.50)
0.74***
(0.17)
4.17
(2.66)
-3.07*
(1.61)
0.17***
(0.06)
-7.48***
(2.64)
0.33
-0.14**
(0.06)
0.03
(0.26)
0.37***
(0.11)
4.56**
(1.77)
-1.35
(0.89)
0.12***
(0.04)
-4.16**
(1.66)
0.27
log 2010
log 2010
GDP Within Emp Within
6 Hours
6 Hours
-0.06
-0.06*
(0.04)
(0.03)
-0.07
-0.07*
(0.04)
(0.04)
0.71***
(0.07)
0.69***
(0.07)
0.05
0.07*
(0.04)
(0.04)
-0.93***
-0.69***
(0.18)
(0.15)
0.38***
0.35***
(0.09)
(0.08)
5.22***
4.36***
(1.60)
(1.28)
-0.19
-0.86
(0.63)
(0.58)
-0.11***
-0.06**
(0.03)
(0.03)
-11.32***
1.03
(2.17)
(1.77)
0.56
0.60
log 2010
Time to
Nearest
Port
0.03
(0.02)
-0.03
(0.02)
0.00***
(0.00)
0.88***
(0.03)
0.22***
(0.08)
-0.07**
(0.04)
-2.11***
(0.63)
0.41
(0.33)
0.09***
(0.02)
0.99*
(0.57)
0.92
First Stage Results – Market Access
1962 Road Rays
1962 Rail Rays
1962/1982 log Full
Market Access
1962/1982 log Dom.
Market Access
1962/1982 log Trade
Market Access
Provincial
Capital
ln 1982 Prefec
Pop
1982 Fraction
HS Graduate
1982 Fraction
Manufacturing
log km to Coast
Constant
R-squared
log 2010
log 2010
log 2010
log 2010 GDP Dom. GDP Trade log 2010
Pop Dom.
GDP Market Market
Market Pop Market Market
Access
Access
Access
Access
Access
0.00
0.00
0.00
0.00
0.00
(0.01)
(0.01)
(0.01)
(0.01)
(0.01)
0.00
0.00
0.00
0.00
0.00
(0.00)
(0.00)
(0.00)
(0.00)
(0.00)
0.21**
0.24***
(0.10)
(0.09)
0.26
-0.69***
0.38**
(0.21)
(0.26)
(0.17)
-0.02
0.80***
-0.13
(0.18)
(0.23)
(0.15)
-0.03
-0.03
-0.03
-0.02
-0.03
(0.02)
(0.02)
(0.02)
(0.02)
(0.02)
0.04
0.04
0.04
0.04
0.04
(0.03)
(0.03)
(0.03)
(0.02)
(0.02)
0.61
0.55
0.64
0.46
0.49
(0.50)
(0.48)
(0.57)
(0.43)
(0.43)
-0.21
-0.21
-0.26
-0.21
-0.20
(0.20)
(0.20)
(0.23)
(0.18)
(0.18)
0.01
0.01
0.01
0.01
0.01
(0.01)
(0.01)
(0.01)
(0.01)
(0.01)
3.13***
2.03
17.66***
5.33***
4.28***
(0.83)
(3.49)
(4.27)
(0.37)
(1.17)
0.06
0.07
0.06
0.07
0.07
log 2010
Pop Trade
Market
Access
0.00
(0.01)
0.00
(0.00)
-0.63***
(0.23)
0.73***
(0.21)
-0.03
(0.02)
0.04
(0.03)
0.65
(0.57)
-0.26
(0.23)
0.01
(0.01)
7.64***
(1.57)
0.06
• MA instruments are a bit weak, especially for domestic and international components
Main Outcomes of Interest
• Prefecture and prefecture city GDP growth
• Prefecture and prefecture city population growth
• Important to take into account that hukou migration restrictions makes it
unlikely to find migration effects
Results for Prefecture GDP
Outcome(s) of Inquiry:
Road Rays
Rail Rays
log 2010 Market
Potential
log 2010 Time to
Coast
Provincial
Capital
ln 1982 Prefec
Pop
1982 Fraction
HS Graduate
1982 Fraction
Manufacturing
log km to Coast
Constant
Exlude 5-7 Nodes
First Stage F
Internal Transport
0.056
(0.048)
0.053
(0.060)
GDP Market Potential
0.13***
(0.033)
0.12***
(0.033)
-0.021
(0.025)
0.15
0.24*** 0.36*** 0.36***
(0.12)
(0.082) (0.077) (0.077)
-0.043
-0.018 -0.13*** -0.13***
(0.065) (0.060) (0.043) (0.042)
-0.33
-0.42
-0.30
-0.36
(0.73)
(0.87)
(0.63)
(0.65)
0.76**
0.60*
0.32
0.27
(0.36)
(0.34)
(0.33)
(0.34)
-0.088*** -0.085*** -0.053*** -0.047***
(0.018) (0.019) (0.015) (0.017)
3.88*** 3.62*** 3.93*** 4.12***
(0.91)
(0.91)
(0.58)
(0.67)
No
No
No
No
20.5
28.7
146
69.4
Employment Market Potential
0.11*** 0.17*** 0.17*** 0.17***
(0.044) (0.036) (0.037) (0.044)
-0.021
-0.016
-0.0069
(0.063)
(0.024) (0.055)
0.26
0.32*** 0.32***
0.28
(0.35)
(0.075) (0.076)
(0.30)
-0.13** -0.13*** -0.13*** -0.13***
(0.049) (0.042) (0.041) (0.047)
0.015
-0.13
-0.16
0.25
(0.75)
(0.63)
(0.64)
(0.74)
0.34
0.44
0.38
0.42
(0.40)
(0.32)
(0.33)
(0.38)
-0.045* -0.060*** -0.054*** -0.053**
(0.024) (0.014) (0.017) (0.025)
4.06*** 2.28*** 2.45*** 2.33**
(0.84)
(0.67)
(0.79)
(0.98)
Yes
No
No
Yes
45.7
105
53.1
40.8
• Estimated effects of market access are consistent with those of market potential
• Results for prefecture cities are also similar, though a bit smaller in magnitude
Results for Prefecture Population
Outcome(s) of Inquiry:
Road Rays
Rail Rays
log 2010 Market
Potential
log 2010 Time to
Coast
Provincial
Capital
ln 1982 Prefec
Pop
1982 Fraction
HS Graduate
1982 Fraction
Manufacturing
log km to Coast
Constant
Exlude 5-7 Nodes
First stage F
Internal Transport
0.0041
(0.022)
0.065**
(0.032)
GDP Market Potential
0.030
(0.022)
0.24*** 0.24*** 0.27***
(0.040) (0.040) (0.045)
-0.087** -0.12*** -0.11***
(0.043) (0.040) (0.043)
-0.011
-0.47
-0.045
(0.24)
(0.33)
(0.26)
0.30*
0.28
0.22
(0.17)
(0.18)
(0.19)
-0.035*** -0.042*** -0.029***
(0.012) (0.013) (0.0084)
1.55** 2.01*** 1.62***
(0.64)
(0.63)
(0.49)
No
No
No
20.5
28.7
146
Employment Market Potential
0.0060
-0.029
0.038*
(0.013) (0.019) (0.021)
-0.064** -0.098*
(0.027) (0.050)
0.28***
0.074
0.27***
(0.041)
(0.27)
(0.040)
-0.12*** -0.049*** -0.12***
(0.036) (0.015) (0.042)
-0.25
-0.37*
-0.0019
(0.23)
(0.22)
(0.26)
0.089
0.28*
0.25
(0.21)
(0.16)
(0.18)
-0.011
0.0020 -0.030***
(0.0092) (0.0099) (0.0088)
2.25*** 1.71*** 1.30***
(0.60)
(0.47)
(0.38)
No
Yes
No
69.4
45.7
105
0.015
(0.013)
-0.062**
(0.027)
0.28***
(0.039)
-0.12***
(0.036)
-0.24
(0.23)
0.088
(0.20)
-0.012
(0.0091)
2.09***
(0.56)
No
53.1
-0.016
(0.017)
-0.085**
(0.043)
0.16
(0.22)
-0.056***
(0.014)
-0.42*
(0.24)
0.23
(0.16)
0.0018
(0.0098)
1.72***
(0.52)
Yes
40.8
• Estimated effects of market access are consistent with those of market potential
• Results for prefecture cities are also similar, though a bit smaller in magnitude
Main Conclusions
• Expansions of highways and railroads in China have promoted regional and urban
growth through mechanisms internal to cities and through improved trade
integration
• Each radial railroad serving a prefecture city allowed it to accommodate an
estimated 7.3 percent more population (6.5 percent more in the full prefecture)
• The estimated elasticity of prefecture GDP with respect to GDP market potential is
0.11 and with respect to employment market potential is 0.17
• No similar effects for prefecture or city population
• The estimated elasticity of prefecture population with respect to travel time to an
international port is about -0.08
• No similar effects for prefecture or city GDP
• May be driven by special economic zones set up to promote foreign investment
• It is important to consider the extent to which these results imply increases in percapita GDP versus its redistribution
Thank You