The Impact of High-Speed Railway Accession on Agricultural Exports: Evidence from Chinese Agriculture-Related Enterprises

Abstract

Although many studies have analyzed the transportation infrastructure effects on economic and trade development, little is known about the relationship between transportation infrastructure and trade in the agricultural sector. We take the opening of China’s high-speed railway (HSR) as a quasi-natural experiment and use multiperiod DID model to explore the impact and mechanism of HSR on agriculture-related enterprises’ exports. The results show that HSR can promote export growth of agriculture-related enterprises by 6.9%, and it will reach 10% in 5 years. However, the effect of HSR on the export of agriculture-related enterprises only exists within 45 km around HSR stations. HSR can reduce information barriers and costs for enterprises to enter the international market by providing transportation convenience and improving market access levels. HSR also offers local areas more transportation advantage compared to other surrounding areas, which in turn makes a siphon effect on export activities. Both these mechanisms are significant within 45 km, and it is extremely obvious for poor transportation areas and enterprises with higher productivity, and the siphon effect is even stronger than market access. Heterogeneity analysis results demonstrate that HSR has different effects for different types of enterprises.

Full text

Research Article
The Impact of High-Speed Railway Accession on Agricultural
Exports: Evidence from Chinese Agriculture-Related Enterprises
Jianjun Zhou ,
1
Xiayang Fan ,
2
and Aizhi Yu
1
1
School of Economics, Central University of Finance and Economics, Beijing 100081, China
2
School of International Trade and Economics, Central University of Finance and Economics, Beijing 100081, China
Correspondence should be addressed to Aizhi Yu; yuaizhi@cufe.edu.cn
Received 6 April 2021; Revised 23 July 2021; Accepted 30 July 2021; Published 13 August 2021
Academic Editor: Bernardo A. Furtado
Copyright ©2021 Jianjun Zhou et al. is is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Although many studies have analyzed the transportation infrastructure effects on economic and trade development, little is
known about the relationship between transportation infrastructure and trade in the agricultural sector. We take the opening of
China’s high-speed railway (HSR) as a quasi-natural experiment and use multiperiod DID model to explore the impact and
mechanism of HSR on agriculture-related enterprises’ exports. e results show that HSR can promote export growth of ag-
riculture-related enterprises by 6.9%, and it will reach 10% in 5 years. However, the effect of HSR on the export of agriculture-
related enterprises only exists within 45 km around HSR stations. HSR can reduce information barriers and costs for enterprises to
enter the international market by providing transportation convenience and improving market access levels. HSR also offers local
areas more transportation advantage compared to other surrounding areas, which in turn makes a siphon effect on export
activities. Both these mechanisms are significant within 45 km, and it is extremely obvious for poor transportation areas and
enterprises with higher productivity, and the siphon effect is even stronger than market access. Heterogeneity analysis results
demonstrate that HSR has different effects for different types of enterprises.
1. Introduction
Agriculture is the foundation of the economy for a country,
and the export of agricultural products is an important
manifestation of the agricultural international competi-
tiveness. Since the 21st century, Chinese agricultural exports
have shown a gradually increasing trend, making China
become the world fifth largest agricultural exporter (Chinese
agricultural exports from 15.449 billion dollars in 2001 in-
crease to 76.989 billion dollars in 2019. e data comes from
the UN Comtrade database). However, Chinese agricultural
exports lack international competitiveness with a lower
market share compared to other countries (in 2019, it
accounted for only 4.89% of the world’s agricultural exports,
which was lower than the EU (39.13%), USA (9.00%),
ASEAN (8.54%), and Brazil (4.97%). e data comes from
the UN Comtrade database and has been sorted and cal-
culated). In China, agricultural products have also shown a
trend of marginalization in the export structure. eir share
of total exports has gradually decreased from 5.81% in 2001
to 3.08% in 2019, while the EU, USA, ASEAN, and Brazil
have maintained their share of agricultural exports stable
and rising trend, keeping an important position in its export
structure (the share of EU agricultural exports in total ex-
ports has gradually increased from 8.12% in 2001 to 9.83% in
2019, the USA has increased from 7.67% to 8.61%, ASEAN
has risen from 7.69% to 9.10%, and Brazil has increased from
27.90% to 34.93%. e data comes from the UN Comtrade
database). e export influence of Chinese agricultural
products is far less than that of developed countries. And,
there is still a considerable gap compared with some de-
veloping countries such as Brazil and ASEAN. More im-
portantly, compared with industrial enterprises, China’s
agricultural enterprises are generally small in scale and low
in technology intensity, in addition to China’s vast territory,
poor traffic, and information conditions in counties and
townships, imposing the export of agricultural products
under high restrictions and barriers to enter the
Hindawi
Complexity
Volume 2021, Article ID 4225671, 19 pages
https://doi.org/10.1155/2021/4225671

international market. In 2017, the Chinese government
proposed a rural revitalization strategy to revitalize the
declining agriculture and rural areas and clearly requested to
promote the prosperity of agricultural and rural industries.
erefore, we need to support the development of agri-
cultural enterprises and encourage agricultural enterprises
to expand the market scope to enter the international trade
market. It can facilitate the export of Chinese agricultural
products and revitalize Chinese agricultural and rural areas.
How to break through geographic restrictions and market
information barriers and lower the thresholds to enter the
international market so that more agricultural enterprises
participate in the export has become a critical issue for
China’s agriculture. e emergence of high-speed railway
(HSR) offers possibilities to solve this problem.
HSR has brought a new growth engine for Chinese
economy, and it also has a huge impact on Chinese agri-
cultural trade. HSR has gone through a fast development in
China since the construction of its first high-speed railway in
2008. By the end of 2019,Chinese HSR operating mileage has
reached 35,000 km, becoming the largest HSR’s country in
the world (the data comes from the 2019 China Statistical
Yearbook). At the same time, the “eight horizontal and eight
vertical” network pattern and trunk lines have basically been
formed, and a series of branch HSR lines have been extended
from trunk lines, further improving Chinese HSR’s network.
As a result, the coverage of HSR in small- and medium-sized
cities and even county areas rather than just large cities has
greatly improved the convenience of transportation in those
areas and stimulated the vitality of local industry. e
downward extension of HSR’s network takes charge of the
agricultural and rural economy within the scope of HSR. Its
space-time compression has greatly narrowed the barriers
and obstacles to communicate with the outside world and
brought huge development opportunities to the agricultural
enterprises in these areas. In county-level and township
areas, agriculture and agricultural product processing in-
dustries account for a large proportion of the industry,
which play a crucial role in the growth of local employment
and income. Under the dual background of Chinese pro-
motion of rural revitalization strategy and the improvement
of HSR’s construction, will HSR become an important force
for agriculture-related enterprises to break through the time-
space limitation and enhance the participation of export? If
HSR can promote the export of agriculture-related enter-
prises, it will raise a series of theoretical and practical
questions: should we explore the degree of HSR’s effect on
the export of enterprises and the impact’s mechanism? Will
the effect of HSR on the exports be affected by the distance
between enterprises and HSR’s stations? Will the policy
effects of HSR on different enterprises be heterogeneous?
e discussion of the above questions will enrich and im-
prove the relevant theories of HSR’s economy and also
expand the application research in Chinese agricultural
trade.
In this paper, we take the opening of Chinese HSR as a
quasi-natural experiment to explore the impact and
mechanisms of HSR on export of agriculture-related en-
terprises. en, we group the distance between enterprises
and HSR stations to further discuss the role of distance
threshold in determining HSR’s effect on enterprises. We use
instrumental variables to deal with endogenous problems
and adopt robustness and placebo tests to verify the reli-
ability of the results. Furthermore, we examine the mech-
anisms from two aspects: market access and siphon effect.
Finally, we analyze the heterogeneous impact of HSR on
different enterprises. To sum up, the research in this paper
will contribute to and further supplement the existing lit-
erature. First of all, although there are many literature
confirming the positive impact of roads, highways, railways,
and other transportation infrastructure on economic growth
and international trade, because HSR is an emerging
transportation infrastructure, there is a lack of literature on
the impact of HSR on international trade. Our paper will
focus on HSR and supplement this part; second, we are
concerned about the development of China’s agriculture and
the impact of China’s HSR in agricultural development.
ere are few literature in this part, but this has very im-
portant practical significance for China’s agricultural de-
velopment; finally, we discuss the mechanism of the impact
of HSR on trade and make up for the lack of previous lit-
erature on mechanism research. en, the rest of the thesis is
organized as follows. Section 2 provides a literature review.
Section 3 offers a theoretical analysis. Section 4 presents
empirical strategy and data. Section 5 presents estimation
results, solution of endogenous problems, robustness
checks, and placebo test. Section 6 discusses the mechanisms
and heterogeneous effects. Section 7 concludes and offers
policy suggestions.
2. Literature Review
Transport improvements can be regarded as a key com-
ponent of regional economic competitiveness. “If you want
to get rich, build roads first” shows the importance of
transportation infrastructure construction for the local
economy [1]. Most studies have fully confirmed the positive
role of transportation infrastructure in regional economic
growth [2–5]. As a fast-reaching transportation, HSR can
operate at a speed of more than 200 km/h, up to and can
reach a maximum of 350 km/h. It significantly breaks the
time-space constraints and has an obvious impact on eco-
nomic activities. Zou et al. explored the impact of the HSR
network on the economic growth of 110 major cities in
China and found that the start of HSR has an apparent
positive effect [6]. Diao et al., respectively, verified the
positive effects of HSR on regional economic growth
according to the changes in fixed asset investment and the
intensity of night light before and after the opening of HSR
[7, 8].
With the deepening of globalization and trade liber-
alization, the impact of transportation infrastructure on
international trade has gradually received greater attention
from researchers. Many studies have shown that the
transportation infrastructure plays a crucial role in reducing
transportation costs and facilitating the growth of trading
activities [9–11]. Countries with large trade value in the
world usually have relatively better domestic transportation
2Complexity

conditions [12–14], and low-level domestic transportation
infrastructure has a restraining effect on the international
trade [15, 16]. In the research on transportation infra-
structure and trade, the traffic objects such as roads and
railways are mainly studied, with relatively less focus on
HSR. e main reason is that HSR is an emerging trans-
portation infrastructure. Only a few countries in the world,
such as China, have a complete HSR’s network. erefore,
the impact of HSR on export has been rarely studied, which
is far from enough and requires further improvement. Xu
et al. provided some research evidences in this field, showing
that the construction of HSR can help promote the growth of
local export [17–19].
Many scholars have discussed how transportation in-
frastructure affects export. e most intuitive conclusion is
that the construction of transportation infrastructure can
reduce the transportation cost of enterprises, and enterprises
can conduct export activities at lower costs, thereby pro-
moting the growth of export [20–23]. However, it is worth
noting that China’s HSR cannot directly reduce trans-
portation costs because it does not directly transport goods.
It enables scholars to explore the mechanism of HSR on
trade from other aspects. Some scholars think that trade
relations depend on the interaction of producers, inter-
mediate traders, and foreign buyers. erefore, close com-
munication between trading partners plays a key role in
information sharing and identification of trade opportuni-
ties [24, 25]. A good transportation infrastructure can in-
crease the frequency of external contacts. It has created an
efficient information communication mechanism to reduce
information barriers and improve trade efficiency. Cosar
and Demir pointed out that the construction of highways has
reduced access barriers to international markets [26]. Some
scholars also believe that HSR plays a very important role in
increasing the frequency of communication between en-
terprises and reducing information costs and barriers.
However, there are not many direct research studies on
trade, most of which are focused on the business activities of
enterprises. Faber pointed out that constructing trans-
portation facilities would connect the central and sur-
rounding cities and reduce trade cost by strengthening
communication and exchanges between regions [27]. is
kind of communication is reflected in improving com-
muting efficiency between regions and exchanging infor-
mation between corporate headquarters and branches [28].
And, it enables enterprises to strengthen their ties with the
outside areas and reduce their market search and business
outsourcing costs by enhancing their matching efficiency
with suppliers, thereby improving the business performance
of enterprises along the route [29].
is paper mainly contributes to the literature in the
following ways. Firstly, more and more literature have
confirmed the positive effects of transportation infrastruc-
ture such as road, highway, and railway on economic growth
and international trade. But there is still lack of research on
HSR in the trade field. So, we will use the export data of
Chinese enterprises to further improve from a micro-
perspective. Secondly, this paper studies the impact of HSR
on the export of agriculture-related enterprises. Previous
studies have rarely involved research on the application of
HSR in the agricultural field. With the improvement of
Chinese HSR’s network and the implementation of rural
revitalization strategy, it is necessary and worthwhile to
study the impact of HSR on agricultural-related industries.
irdly, we have discussed the impact of the distance be-
tween enterprises and HSR’s station on the export and
confirmed that the opening of HSR can only affect the export
of enterprises within 45 km, with no significant effect beyond
45 km. is is a novel point of view compared to previous
studies. Finally, the previous literature has not sufficiently
discussed the mechanism of HSR on trade. We have con-
sidered the mechanism of HSR on trade from two aspects:
market access and siphonic effect, which have effectively
made up for the deficiencies of the previous literature on the
mechanism. China’s vast territory provides extremely rich
agricultural resources, and plenty of agriculture-related
enterprises use agricultural products as their main raw
materials or final products, which provide rich samples for
this research. is paper is significant for China to attach
great importance to agriculture and rural areas, and it also
has significant referential effects for other developing
countries with the attempt to establish rapid transportation
infrastructure similar to HSR to promote agricultural export.
3. Theoretical Background and
Analysis Framework
We explore the effect of HSR on export under the following
theoretical background. With reference to the methods of
Grossman et al. [30, 31], we assume that the export products
of agriculture-related enterprises need to be made by several
manufacturers in multiple regions, from raw material ac-
quisition, product processing, and transportation to external
sales. is process is restricted by the cost of information
such as communication between the two regions. e
substantial improvement of transportation infrastructure
made by HSR can greatly improve the local market access
level. Local agriculture-related enterprises, especially those
around HSR’s stations, can take full advantage of the
transportation superiorities brought by HSR, quickly reach
other regions, and establish more frequent contacts with
producers, middlemen, and traders in other regions. It can
help agriculture-related enterprises fully understand the
market information and find more production orders and
trade opportunities. erefore, this passenger-oriented
transportation mainly improves the market access level and
lowers the threshold of entering the local market. It is
conductive to the frequent face-to-face communication
between local enterprises and the outside world to drive the
growth of economic and trade activities.
HSR can improve the local market access level and then
guide economic and trade activities in surrounding regions
without HSR flow to HSR’s location. It is easy to cause trade
competition between different regions, resulting in a
siphonic effect. Krugman [12] used “center-periphery”
model to analyze economical activities’ spatial location and
believed that economic activities tend to gather from pe-
riphery to central area, which has a siphonic effect on
Complexity 3

economic and trade activities in peripheral areas. HSR will
increase spatial mobility of the resources and the clustering
from non-HSR regions to HSR’s regions and enable local
agriculture-related enterprises to gain benefits in trade
competition among similar enterprises in surrounding areas.
Since agriculture-related enterprises mainly export low-end
and homogenized products, export trade activities have
strong regional substitutability. HSR helps local enterprises
to compete for more trade opportunities and export orders,
forming a siphonic effect for enterprises in surrounding
regions. While promoting the export growth of agriculture-
related enterprises in HSR’s regions, it has also led to a
decrease in exports of enterprises without HSR.
Furthermore, we take the mechanism of market access as
an example for the theoretical derivation of export. In in-
ternational trade, we treat all countries as our own country
and foreign countries. Our country is composed of Nre-
gions, and each has a lot of agriculture-related enterprises.
ere is also a trade relationship between agriculture-related
enterprises in each region and foreign countries. For easy
distinction, we mark the starting point of the export as
region aand all destination countries of export as country b.
3.1. Consumer Preferences. We assume that consumers in
country bconsume a series of differentiated products iand
have a standard CES preference for product i. e utility
function is
Ub�􏽚Ω
0
xb(i)(σ−1)/σdi
􏼠 􏼡σ/(σ−1).(1)
In formula (1), Ωrepresents the set of products available
for consumers in country b,σrepresents the elasticity of
mutual substitution between products, and σ>0, and xk(i)
represents the consumption of product iby consumers in
country b, which is constrained by the income level of
consumers in country b:
yb�􏽚Ω
0
pb(i)xb(i)di. (2)
In formula (2), pb(i)represents price of product iin
country band ybrepresents the per capita income level of
country b.
3.2. Production Technology Level. We assume that produc-
tion factors of each region include land (L), labor (H), and
capital (K). e production function is in the form of
Cobb–Douglas:
Xa(i) � za(i)La(i)αHa(i)cKa(i)1−α−c,(3)
MCa(i) � qα
awc
ar1−α−c
a
za(i).(4)
In formula (3), za(i)represents average productivity,
MCa(i)represents marginal cost of product iin the region a,
qa,wa, and ra, respectively, indicate the factor return rates of
land, labor, and capital, and average productivity za(i)obeys
following distribution:
Fa(z) � Pr Za≤z
 􏼁�exp −Aaz−θ
􏼐 􏼑.(5)
In formula (5), θrepresents the change in productivity
within region aand Aarepresents the technological level of
region a.
At the enterprise level, we learn from Melitz and assume
that the production of agriculture-related enterprises has
increasing returns to scale, the products produced have
subtle differences, and there is heterogeneity among en-
terprises [32]. erefore, productivity φof agriculture-re-
lated enterprises obeys following distribution:
G(φ) � 1−φc.(6)
In formula (6), φ≥1 for an agriculture-related enterprise
mwith a productivity of φ, the marginal cost of production is
(mcs/φ), and mcsrepresents marginal cost of lowest-pro-
ductivity firm s, so the higher a company’s productivity is,
the lower marginal cost it will face.
3.3. Trade Cost and Enterprise Export. We use a simple
“iceberg cost” to describe trade cost, assuming that the trade
cost between aand bis τab, and the trade cost between aand
bis symmetrical, that is, τab �τba,τab �τba >1. If product i
is produced in region aand directly supplied to region a, the
price of product iis paa(i). But if product iis produced in
region aand then sold to country bthrough international
trade, the price of the product iis pab(i); these two prices
satisfy the following relationship:
pab(i) � τab ×paa (i) � τab ×MCa(i) � τab ×qα
awc
ar1−α−c
a
za(i).
(7)
For ordinary consumers, when they make a commodity
purchase decision, they are sensitive to its price, and they
usually choose commodity with the lowest price, so goods’
price is often affected by productivity distribution.
From Eaton and Kortum, we can get two important
conclusions [33]. Firstly, the price index of trade destination
country bsatisfies
Pb
 􏼁−θ�κ1􏽘
a
Aaqα
awc
ar1−α−c
a
􏼐 􏼑−θτ−θ
ab
􏼔 􏼕≡CMAb.(8)
In formula (8), we define CMAbas the market access of
consumers in country b, which reflects low-priced products’
availability in market for consumers in country b. e other
important conclusion is the total value of products sold from
region ato country b, as shown below:
Xab �κ1Aaqα
awc
ar1−α−c
a
􏼐 􏼑Ybτ−θ
ab CMA−1
b.(9)
In formula (9), we can see that the total value of products
Xab sold by region ato country bis a standard gravitational
equation. When the target country’s trade cost decreases, the
export trade from region ato country bwill increase;
otherwise, export trade will decrease.
4Complexity

3.4. Market Access and Enterprise Export. Since we treat all
export trading countries as country b, we can sum up
formula (9) to get
Xa�􏽘
b
Xab �κ1Aaqα
awc
ar1−α−c
a
􏼐 􏼑−θ􏽘
b
τ−θ
ab CMA−1
bYb
􏼐 􏼑.
(10)
In formula (10), Xais the total income of region a
representing whole exports to all foreign countries, that is,
the total amount of exports. We define the enterprise market
access level of region aas follows:
FMAa≡􏽘
b
τ−θ
ab CMA−1
bYb.(11)
In formula (11), enterprise’s market access is composed
of the market size Yb, the consumer market access CMAbof
the destination country, and trade cost τab between region a
and country b. Market size and level of market competition
in country bare often not affected by region a, so the market
access level of enterprises in region alargely depends on
trade cost τab between these two places.
Because the trade cost between aand bis symmetrical,
that is, τab �τba, according to Donaldson and Hornbeck, it
can be obtained FMAa�λCMAaand FMAb�λCMAb,
λ>0. It shows that, for the same region, there is a linear
correlation between the enterprise market access and the
consumer market access, and this linear correlation feature λ
will not affect the analysis, so we can get
FMAa�λCMAa�MAa,(12)
MAa�λ􏽘
b
τ−θ
ab MA−1
bYb.(13)
We put formula (13) into (10), and we can get
Xa�κ2Aaqα
awc
ar1−α−c
a
􏼐 􏼑−θMAa.(14)
Substituting the factor income of land, labor, and capital
into formula (14), we can get
Xa�κ3Aa
 􏼁1/(1+θα+θc)α
La
􏼠 􏼡− (θα/1+θα+θc)c
Ha
􏼠 􏼡− (cα/1+θα+θc)1−α−c
Ka
􏼠 􏼡− ((1−α−c)α/1+θα+θc)MA(1+θ(1+α+c))/θ(1+θα+θc)
a.(15)
In formulas (13)–(15), we can see that when the cost of
trade between aand bdecreases, the market access level of
region acan be greatly improved, further increasing the
income level brought by export trade. With the size of the
destination market and the degree of market competition
unchanged, reducing trade cost between these two places is a
crucial way to ameliorate market access and promote export
trade.
4. Methods and Materials
4.1. Empirical Model Setting. Considering that different
regions open HSR in different times, we learn from the
methods of Lin and Qin and use a multiperiod DID model to
examine the effect of HSR on export of agriculture-related
enterprises [34, 35]. In the DID model, we take the regions
where the HSR has not been opened as the control group and
the regions where the HSR has been opened as the exper-
imental group and obtain the policy effect of the HSR’s
opening through two differences. e benchmark model
adopts following settings:
ln exportict �β1HSRct +β2Eict +β3Cct +ξi+ξpt +ξnt +μict.
(16)
In formula (16), ln exportict represents agriculture-re-
lated enterprise’s export value and HSRct indicates whether
this region chas opened HSR. When HSR is opened,
HSRct �1; otherwise, HSRct �0. β1is the estimated coef-
ficient of HSR. In order to obtain the net effect of HSR’s
opening on the export of enterprises, we need to control
other factors that affect the export of enterprises. It mainly
comes from two aspects: on the one hand, it comes from the
enterprise itself, and some enterprise’s own factors will also
affect the export. Eict are control variables at the enterprise
level, including variables that measure individual charac-
teristics of agriculture-related enterprises to control their
impact on exports such as the size of enterprise (lnsize); the
larger the enterprise scale, the stronger the export capacity.
e operating time of enterprise (lnage): as the operating
time has become longer, the business and trade relationships
of the enterprise have gradually stabilized, and the enterprise
has also passed the dangerous period of survival. It is more
likely to explore the higher risk international market, which
will help the enterprise’s export growth. e labor pro-
ductivity (lnlarborate): Melitz points out that enterprises
with higher productivity are more likely to engage in export
activities. erefore, enterprise productivity has a significant
positive impact on exports. Other factors that affect the
export of enterprise are the external trade dependence
(open) and financial liquidity (finance). On the other hand,
some region characteristics will also affect the export ac-
tivities of enterprises. Cct are control variables at the regional
level that affect export such as the level of road traffic
(lnroad); good road conditions are conducive to product
transportation, reduce transportation and trade costs, and
then promote enterprise exports. e gross domestic
product of agriculture (lnagriculture): a region with a higher
agricultural production value will make the agricultural
enterprises in that region more likely to export. In addition,
economic development (lnpgdp), total population (lnpo-
pulation), and Internet development (lninternet) will also
affect the exports of enterprise. erefore, we need to add
these control variables to the empirical model. For some
Complexity 5

fixed effect choices, we use a fixed effect model to analyze the
impact of HSR on enterprise exports, so we need to control
the fixed effects at the enterprise level. ξiis a enterprise’s
fixed effect, which is used to control factors that enterprise
does not change over time. Since enterprise’s location is
fixed, when we control enterprise’s fixed effect, regional fixed
effect will be controlled accordingly. Finally, in the industry
and provincial level, there are some unobservable time
trends, which will also affect the export of enterprises. For
example, if some industries develop rapidly, the export of
enterprises in these industries will be significantly faster than
that of enterprises in other industries. erefore, we control
the industry-time fixed effect and province-time fixed effect
in the model setting. ξnt represents industry-time fixed effect,
ξpt represents province-time fixed effect, which is used to
control time trend at the industry and provincial level, and
μict represents the random disturbance item.
4.2. Parallel Trend Test. Due to the inconsistency of HSR’s
opening time in different regions, it is impossible to directly
obtain a parallel trend of policy effects. erefore, we learn
the event analysis method from Beck et al. [36] and add
dummy variables before and after the policy on the basis of
formula (16):
ln exportict �􏽘
10
m�1
βmBFHSRc,t−m+􏽘
6
n�0
βnAFHSRc,t+n
+β2Eict +β3Cct +ξi+ξpt +ξnt +μict.
(17)
In formula (17), BFHSRc,t−mrepresents the myears
before HSR’s opening and AFHSRc,t+nrepresents the nyears
after HSR’s opening. When the coefficient βmis close to 0, it
indicates that, before the opening of HSR, there is no sig-
nificant difference in exports of agriculture-related enter-
prises between the experimental and control group. When
the coefficient βnis significantly different from 0, it is means
policy effect is very obvious. e results are shown in Fig-
ure 1: HSR has brought a significant and continuous in-
creasing for agriculture-related enterprises’ export.
4.3. e Distance between Agriculture-Related Enterprises and
HSR Stations. e reason why the problem of distance is
introduced to the analysis of this article is mainly due to the
differences in city form and geographic location of enter-
prises, which are rarely considered in previous studies. ey
treat different cities as homogeneous, regardless of the geo-
graphical distribution of enterprises in the city and the
resulting distance issues. In China, the differences between
prefecture-level cities are very obvious. Some prefecture-level
cities have a huge area and are composed of many counties.
e distance from east to west and from south to north is very
long. Some prefecture-level cities are very small, consisting of
only municipal districts and a few counties. In addition, the
construction sites of China’s HSR stations are also quite
dissimilar. Some are rebuilt from the original railway stations,
and these HSR stations are often closed to the city center.
Some cities elect to build HSR stations in the suburbs due to
land rent; these HSR stations are often far from the city center
and industrial parks. at makes the distance between en-
terprises and HSR stations very different. Some enterprises are
close to the HSR station and can enjoy the convenience of
transportation brought by the opening of the HSR, thereby
helping enterprises make better use of the HSR to carry out
economic and trade activities and drive the growth of exports.
Enterprises that are far away from the HSR station take longer
time to reach the HSR station and use the HSR less frequently
so that the opening of HSR has almost no impact on the
exports of such enterprises. We take a sample of agriculture-
related enterprises in regions where HSR was opened in 2013
as an example and analyze the spatial distance distribution
between agriculture-related enterprises and HSR stations (the
radius is roughly calculated based on the area of the pre-
fecture-level city’s administrative and district. e calculation
formula is (s/π)
(1/2)
, where sis the area and πis the ratio of the
circumference of a circle to its diameter. Among them, the
prefecture-level city’s district is generally the central city
where the prefecture-level city’s government is located, and it
is also a city in a narrow sense). As shown in Figure 2, taking
15 km as an interval, we can see that the distance between
enterprises and HSR stations is mostly within 105 km, and the
number of enterprises in 15–30 km interval is the largest.
ere are many enterprises within 45 km, and the apex of the
normal distribution curve is also in 30–45 km interval. At the
same time, we also discussed the radius of prefecture-level
city’s administrative area, the radius of prefecture-level city’s
districts, and the distance from HSR stations to a city’s center.
As shown in Figures 3–5, a large number of prefecture-level
city’s administrative area’s radius in China are within 120 km,
with the most in the range of 50–70 km, while the radius of
prefecture-level city’s districts is mostly within 20 km. e
distance from the HSR station to city center is also mostly
concentrated within 25 km.
4.4. Data Source and Variable Description. e first part is
HSR data of prefecture-level cities, which is mainly from
China Railway Corporation website, China Railway
–0.2
–0.1
0
0.1
0.2
0.3
0.4
Coefficient
t-10 t-9 t-8 t-7 t-6 t-5 t-4 t-3 t-2 t-1 t t+1 t+2 t+3 t+4 t+5
Period
Figure 1: Parallel trend test on the export of agriculture-related
enterprises.
6Complexity

Yearbook, and 12306 website. We manually collect HSR
lines and the opening time of each station from 2000 to 2013.
We use the Baidu Map API to obtain the latitude and
longitude coordinates of agriculture-related enterprises and
HSR stations along the line and calculate the straight dis-
tance between agriculture-related enterprises and the nearest
HSR station.
e second part is the microdata of agriculture-related
enterprises, which mainly comes from two sources. e first is
0
0.005
0.01
0.015
0.02
Density
0 50 100 150
km
Figure 2: Distance between enterprises and HSR stations.
0
0.005
0.01
0.015
0.02
Density
0 50 100 150 200 250
km
Figure 3: e radius of prefecture-level city’s area.
0
0.05
0.1
0.15
Density
010203040
km
Figure 4: e radius of prefecture-level city’s districts.
0
0.02
0.04
0.06
0.08
0.1
Density
0 5 10 15 20 25
km
Figure 5: e distance from HSR stations to the city’s center.
Complexity 7

from China Industrial Enterprise Database, which provides
detailed enterprise information, including the year of es-
tablishment, total industrial output value, industrial sales
output value, and industry. e second is the export data of
enterprises, which comes from the Chinese Customs Data-
base. e database records detailed trade information of
enterprises, including export products, export values, quan-
tities, prices, and export destination countries. We use the
method of Brandt et al. to match the microenterprise data of
these two databases to get export information of Chinese
industrial enterprises and further obtain samples of agri-
culture-related enterprises from matching data [37]. We
adopt two methods to identify whether the enterprise is an
agriculture-related enterprise: e first is based on industry
attributes in the Chinese industrial enterprise database.
According to the classification method of the soft science
research group of the Ministry Agriculture China, the in-
dustry category codes in Chinese industrial enterprise data-
base are 13–23 and 29, including agricultural and sideline
food processing, and food manufacturing belongs to agri-
cultural product processing industries. e second depends
on HS classification code in Chinese Customs Database.
Export products of HS01–HS24 are usually classified as ag-
ricultural products, and we also classify enterprises that ex-
port such goods as agriculture-related enterprises. After
screening export samples, we add them to the enterprise level
by year. en, we refer to methods of Brandt et al. [37, 38] and
exclude samples with fewer than 8 employees, industrial
output value, total assets, fixed assets, and sales revenue from
main business with zero or missing, as well as samples that do
not conform to GAAP, including enterprises whose both
current assets and fixed assets are greater than total assets.
e third part is economic data at the regional level,
which is mainly from the China Regional Statistical Year-
book and the National Bureau of Statistics of China, in-
cluding economic development, road traffic conditions,
agricultural output level, population, and informatization in
each region. Descriptive statistics: mean, standard deviation,
minimum, median, and maximum values for each variable
are shown in Table 1.
5. Results and Discussion
5.1. Benchmark Regression Results. Under the controlling of
firm fixed effects, province-time fixed effects, and industry-
time fixed effects, we sequentially add control variables for
regression (because provincial capitals and municipalities
have obvious political advantages in the HSR construction
planning, in order to eliminate the estimation bias on the
model results, the sample of enterprises located in provincial
capitals and municipalities will be deleted). e results are
shown in Table 2. With the successive addition of control
variables, the coefficients of HSR are all significantly positive
at the 1% level, indicating that HSR can significantly pro-
mote agriculture-related enterprises’ export growth. In
column (10), the result shows that HSR will increase the
export growth of agriculture-related enterprises by about
6.9%, compared with enterprises in regions without HSR.
When more and more small- and medium-sized cities and
counties open HSR, enterprises and individuals in these
regions will enjoy the policy dividends brought by HSR.
Especially for Chinese agricultural and rural areas, HSR
provides an opportunity to conduct external communica-
tion and exchanges in the Chinese agricultural system. It will
reduce the threshold and cost to accept peripheral infor-
mation and expand market search radius for agriculture-
related enterprises. At the same time, HSR guides economic
and trade activities to gather in the region of HSR’s location
through passenger transportation, strengthening the local
advantage in economic and trade competition with non-
HSR regions. More importantly, HSR can encourage agri-
culture-related enterprises to participate in international
competition and promote their products to the international
market.
For the control variables, it is important to control the
impact of other factors on the export of the enterprise so that
we can accurately identify the impact of HSR’s opening on
enterprise exports. From the perspective of enterprise, the
age, scale, export dependence, financing restrictions, and
labor productivity of the enterprise will have an impact on
the export. And, from the perspective city, economic de-
velopment, road traffic conditions, agricultural output value,
total population, and Internet informatization level will also
have an impact on the export of agriculture-related enter-
prises. At the enterprise level, we can find that the coefficient
estimates of control variables have a very significant positive
impact on the export of agriculture-related enterprises. At
the regional level, economic development has a critical
negative impact on the export of agriculture-related en-
terprises. at is mainly because the region with higher
economic development level has lower agriculture pro-
portion. Correspondingly, the coefficients of the agricultural
output value on the exports are significantly positive at the
level of 5%. If a region has a higher agricultural output value,
it can provide more abundant agricultural resources as raw
materials for enterprise production. e coefficients of road
traffic conditions on exports are also remarkable and pos-
itive. Ameliorating road traffic conditions can significantly
reduce transportation time and cost and promote the out-
ward of agricultural products. e coefficients of population
density are not significant, while the level of Internet
informatization has a positive impact on exports at the 10%
level. It shows that, to a certain extent, the improvement of
informatization can help enterprises to obtain market in-
formation, which is beneficial to export behavior. e co-
efficients of main control variables are statistically significant
and in line with expectations and economic principles.
5.2. Dynamic Effect of HSR on Exports. In order to explore
the dynamic impact of HSR on export, we examine the
coefficient of HSR lag term on exports. As shown in columns
(1)–(5) in Table 3, HSR has a clear dynamic impact on the
exports of agriculture-related enterprises. It can be observed
that, in 1–5 years after, HSR still has a positive effect in
promoting export growth of local agriculture-related en-
terprises. Especially in the fourth to fifth year, the policy
effect of HSR has been significantly improved, which can
8Complexity

increase export growth by about 10%. In addition, we
multiply HSR and time dummy variable after opening as the
explained variable and add it to the model. As shown in
column (6) in Table 3, in 2008, when HSR was first put into
operation, the export growth of agriculture-related enter-
prises was not obvious. After 2009, the coefficient of the
interaction term was very positive and gradually increased,
and it was 0.093 and 0.110in 2012 and 2013, respectively, and
both were significant at the 1% level, which was consistent
with the regression results of the lag term. e main reason is
that there are only 3 HSR lines opened in 2008, and the lines
have very short mileages and few stations (the 3 HSR lines is
the Jing-Jin intercity railway from Beijing to Tianjin, the He-
Ning section of the Ning-Rong railway from Hefei to
Nanjing, and the Jiao-Ji railway from Jinan to Qingdao).
Since 2009, the lines of HSR have gradually increased. China
has successively opened many main lines such as the Beijing-
Guangzhou HSR and the Beijing-Shanghai HSR. As the HSR
network has gradually improved, a large number of regions
along the lines have been included in the HSR network,
Table 1: Descriptive statistics of each variable.
Variable Variable explanation Mean SD Min p50 Max
lnexport Logarithm of export value 14.099 1.902 8.189 14.381 17.771
HSR Whether to open high-speed railway 0.253 0.435 0.000 0.000 1.000
lnage Enterprise age, logarithm of opening time 2.083 0.648 0.000 2.197 3.761
lnsize Enterprise size, logarithm of total assets 15.290 1.339 12.549 15.186 19.002
Open Export dependence, export delivery value divided by total industrial output
value∗100% 48.420 42.243 0.000 46.928 123.246
Finance Financing constraints, current assets divided by total assets∗100% 55.784 26.114 1.513 58.833 97.832
lnlarborate Labor productivity, logarithm of per capita gross industrial output 10.377 1.017 8.053 10.334 13.070
lnpgdp e level of economic development, logarithm of per capita GDP 8.549 0.783 5.892 8.650 9.875
lnroad Logarithm of local road density 0.708 0.223 0.186 0.710 1.158
lnagriculture Logarithm of the agriculture gross domestic product 18.850 0.828 16.287 18.969 20.366
lnpopulation Logarithm of local population density 6.632 0.769 4.212 6.563 8.375
lninternet Internet penetration, number of computers per 100 households 13.425 1.033 10.608 13.505 15.512
Table 2: e impact of HSR’s opening on the export: benchmark model results.
Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
HSR 0.062∗∗∗ 0.076∗∗∗ 0.076∗∗∗ 0.074∗∗∗ 0.082∗∗∗ 0.068∗∗∗ 0.063∗∗∗ 0.068∗∗∗ 0.068∗∗∗ 0.069∗∗∗
(3.67) (4.56) (4.60) (4.51) (4.92) (4.10) (3.83) (4.09) (4.02) (4.10)
lnage 0.318∗∗∗ 0.235∗∗∗ 0.207∗∗∗ 0.204∗∗∗ 0.195∗∗∗ 0.196∗∗∗ 0.195∗∗∗ 0.195∗∗∗ 0.195∗∗∗ 0.195∗∗∗
(19.61) (14.77) (13.12) (13.00) (12.49) (12.53) (12.45) (12.41) (12.41) (12.37)
lnsize 0.345∗∗∗ 0.347∗∗∗ 0.351∗∗∗ 0.320∗∗∗ 0.320∗∗∗ 0.321∗∗∗ 0.321∗∗∗ 0.321∗∗∗ 0.320∗∗∗
(37.50) (37.63) (38.12) (34.61) (34.69) (34.74) (34.74) (34.71) (34.44)
Open 0.006∗∗∗ 0.006∗∗∗ 0.006∗∗∗ 0.006∗∗∗ 0.006∗∗∗ 0.006∗∗∗ 0.006∗∗∗ 0.006∗∗∗
(46.76) (45.79) (47.06) (46.99) (46.99) (46.95) (46.94) (46.62)
Finance 0.002∗∗∗ 0.002∗∗∗ 0.002∗∗∗ 0.002∗∗∗ 0.002∗∗∗ 0.002∗∗∗ 0.002∗∗∗
(10.53) (10.63) (10.53) (10.49) (10.52) (10.52) (10.51)
lnlarborate 0.151∗∗∗ 0.152∗∗∗ 0.153∗∗∗ 0.152∗∗∗ 0.152∗∗∗ 0.152∗∗∗
(23.36) (23.57) (23.62) (23.54) (23.53) (23.43)
lnpgdp −0.192∗∗∗ −0.203∗∗∗ −0.206∗∗∗ −0.206∗∗∗ −0.209∗∗∗
(−3.45) (−3.65) (−3.69) (−3.54) (−3.60)
lnroad 0.184∗∗∗ 0.180∗∗∗ 0.180∗∗∗ 0.165∗∗
(2.70) (2.63) (2.63) (2.42)
lnagriculture 0.098∗∗ 0.098∗∗ 0.102∗∗
(2.33) (2.32) (2.38)
lnpopulation 0.001 0.020
(0.00) (0.16)
lninternet 0.023∗
(1.66)
_cons 12.505∗∗∗ 7.435∗∗∗ 7.000∗∗∗ 6.885∗∗∗ 5.771∗∗∗ 7.492∗∗∗ 7.465∗∗∗ 5.652∗∗∗ 5.647∗∗∗ 5.212∗∗∗
(60.08) (29.51) (27.00) (26.52) (21.51) (13.26) (13.19) (5.87) (3.94) (3.59)
Individual FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
N179236 179228 170161 170161 169520 169520 169520 169319 169319 167886
Adj. R-sq 0.086 0.107 0.135 0.136 0.143 0.143 0.143 0.143 0.143 0.143
Note. ∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
Complexity 9

which has strengthened communication and connection
between inland regions along HSR lines and the developed
coastal areas or port cities. As a result, the policy effect of
HSR on local enterprises’ exports has gradually increased,
which is consistent with the results in Figure 1. It is fore-
seeable that the HSR will play an increasingly important role
in the export process of Chinese agriculture-related enter-
prises in the future.
5.3. Distance reshold for the Impact of HSR on Exports.
e policy effects of HSR may be affected by the distance
between agriculture-related enterprises and HSR stations.
We calculated the straight distance between each agricul-
ture-related enterprise and the nearest HSR station, taking
15 kilometers as an interval and dividing it into 0∼15, 15∼30,
30∼45, 45∼60, 60∼75, 75∼90, and 90∼105 km. en, we
examine the HSR’s coefficients of different distance
thresholds on exports. As shown in Table 4, when the
distance is within 45 km, HSR will significantly enhance
export of agriculture-related enterprises. It will promote the
export growths about 7.9% within 15 km and 10% between
15 to 45 km especially. When the distance exceeds 45 km,
HSR will no longer have apparent impact on exports.
erefore, we can see that HSR has a very obvious spatial
distance threshold about 45 km for exports.
Figures 2–5 show the distance between agriculture-re-
lated enterprise and HSR’s station, the radius of prefecture-
level city’s administrative region and district, and the dis-
tance from the HSR station to the city center. We can find
that the threshold of 45 km is roughly equivalent to 2 times
the radius of prefecture-level city’s district and 2 times the
distance between the HSR station and the city center. We use
a simplified diagram to analyze the influence of the distance
threshold, as shown in Figure 6: the circle Nrepresents the
range of the city’s districts, the region out of Mrepresents the
county, township, and rural area, Orepresents the city
center, ON represents the radius of city’s districts, and the
distance of ON is 20 km. According to the distance between
the HSR’s station and the city center, we first assume that the
HSR’s station is located in the center of city, as expressed by
point S. It can affect the surrounding agriculture-related
enterprises within 45 kilometers (within the circle M). e
enterprise uwithin the scope can conveniently use HSR to
carry out business communication and trade activities, and
foreign enterprises can also easily reach the enterprise u
through the HSR station, thereby promoting trade oppor-
tunities and export orders of enterprise u. e enterprises v
outside this range are far away from the HSR’s station, and it
takes longer time to reach the HSR’s station, leading to less
frequent economic and trade activities through HSR to other
regions and reducing the willingness and subjective initiative
to conduct market search and expansion. It is unfavorable
for the enterprise vto carry out economic and trade ac-
tivities. When the HSR’s station is located outside the circle
N, it can also cover the entire districts of the city, and the
result is consistent with the former.
5.4. Resolution of Endogenous Problems. In this study, the
endogenous problem mainly comes from the nonrandom
nature of HSR construction. Regions with better economic
development conditions tend to have more possibility of
opening HSR, and they may have more export trade ac-
tivities. We use the instrumental variable regression method
to do further endogeneity processing [39]. Firstly, we learn
from the method of Faber and use the “least-cost path-
spanning tree network” as an IV for the opening of HSR
[27]. Secondly, we use China’s railway lines in 1961 as an IV.
A historical railway line of 1961 has reference meaning for
the designing of HSR. erefore, historical lines have a high
correlation with HSR lines, and it is not related to other
factors that affect enterprises’ export, meeting the exogenous
assumption. Finally, we also use the railway passenger
Table 3: e impact of HSR on the exports of agriculture-related enterprises: dynamic effects.
Variable (1) (2) (3) (4) (5) (6)
L.HSR 0.090∗∗∗ −0.028
(4.75) (−0.72)
L2.HSR 0.087∗∗∗ 0.061∗∗
(4.28) (2.33)
L3.HSR 0.080∗∗∗ 0.061∗∗∗
(3.28) (2.65)
L4.HSR 0.107∗∗∗ 0.065∗∗∗
(3.28) (2.99)
L5.HSR 0.106∗0.093∗∗∗
(1.80) (3.95)
HSR ∗year 0.110∗∗∗
(3.83)
Control variables Yes Yes Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes Yes Yes
N118238 89899 67320 50387 37828 167886
Adj. R-sq 0.092 0.076 0.074 0.075 0.075 0.143
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively. And, the tstatistic value is in parentheses. e results of other control
variables are not displayed in the table, the same as below.
10 Complexity

volume of each region in 1998 as an IV. HSR needs to give
priority to regions with more transport demand and satisfy
correlation assumptions. Moreover, it is unlikely that his-
torical traffic and transportation pattern are related to other
factors affecting current enterprise’s export, which satisfies
the exogenous assumption.
e results of IV regression are shown in Table 5: the
coefficients of HSR in IV regressions are 0.394, 0.335, and
0.371, which are all very positive at the 1% level, and co-
efficients of these three are relatively close. At the same time,
compared with coefficient of 0.069 in the benchmark re-
gression model, the coefficients of IV regression are only
expanded by 4–6 times. Jiang analyzed the papers using IV
regression in the top financial journals and found that IV
estimation expanded the coefficient by an average of 9 times
[40]. e expansion of HSR’s coefficients in this study is in
an acceptable range, indicating that the estimation of IV is
more reliable. In addition, the first-stage regression coeffi-
cients of IV for HSR are, respectively, 0.050, 0.030, and 0.023,
which are significantly positive at the 1% level. e Klei-
bergen-Paap Fstatistic far exceeds the first-stage empirical
value of 10 [41], so the hypothesis of weak instrumental
variables can be rejected, which fully reflects the effectiveness
of IV.
5.5. Robustness Test. We will adopt the following methods to
verify the robustness of the estimation results: (1) we replace
the explained variables and use the total export volume and
per capita export delivery value instead of export value to
perform model regression. (2) We will add samples of en-
terprises in the municipalities and provincial capitals. (3)
Keep data from 2008–2013 and shorten the sample period.
(4) Keep odd and even year data separately. (5). Extending
the sample period to 2016, use the export data of agricultural
products from the 2000–2016 for model regression (we will
not be able to control the influencing factors at the enterprise
level by using the export data of agricultural products in
Chinese Customs Database from 2000 to 2016. Only the
influencing factors at the city level can be controlled). e
robust results are shown in Table 6; all coefficient are sig-
nificantly positive at the 1% level and relatively close to the
0.069 obtained in benchmark model results (the average
value of seven coefficient estimates is 0.063), which can
explain why the empirical results obtained by model esti-
mation are robust and reliable.
After checking the robustness of the benchmark re-
gression results, we need to further verify the robustness
affected by distance threshold of HSR to agriculture-re-
lated enterprises’ export. In Table 4, we conducted a
segmented regression with 15 km as an interval. ere-
fore, we first change the interval and use the 14 and 16 km
adjacent to 15 km for segmentation. Secondly, we use half
of the 45 km interval at 22.5 km, then use the 22 and 23 km
adjacent to 22.5 km for segmentation, and verify the ro-
bustness of the distance threshold through squeeze the-
orem. e results are shown in Table 7; in columns (1)-(2),
the coefficients of HSR are significantly positive within 42
and 48 km, indicating that HSR has an apparent effect on
increasing the export of agriculture-related enterprises
Table 4: e impact of HSR on the export: different geographical distance intervals.
Variable [0,15) [15,30) [30,45) [45,60) [60,75) [75,90) [90,105)
(1) (2) (3) (4) (5) (6) (7)
HSR 0.079∗∗ 0.105∗∗∗ 0.100∗∗ 0.006 0.074 0.138 −0.164
(1.97) (2.90) (2.39) (0.13) (1.01) (1.33) (−0.95)
Control variables Yes Yes Yes Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes Yes Yes Yes
N34360 36949 30203 24169 13113 8613 5305
Adj. R-sq 0.153 0.171 0.163 0.192 0.143 0.112 0.159
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
N
M
uv
OS
N
M
Su
vO
Figure 6: Distribution of the city center, HSR’s station, and enterprises.
Complexity 11

within this distance. In columns (3)–(5), the coefficient
estimates are very positive at the level of 5% within 45 km,
indicating that the estimation of spatial distance threshold
is robustness and reliability. We have further changed the
division of distance intervals. For example, we divided by
13, 17, 21, and 24 km and performed model regression,
which had no effect on the robustness of results. e effect
of HSR on the export of agricultural enterprises is about
45 km, and once exceeding this distance, HSR will no
longer have an impact on the export of agricultural
enterprises.
5.6. Placebo Test. We conduct a placebo test on the re-
gression results of the benchmark model. is placebo test
has two main parts: firstly, we make the policy shock of HSR
on agriculture-related enterprises in specific regions become
random (the policy shock is generated randomly by a
computer) and estimate the impact of HSR on agricultural
enterprises’ exports. en, we repeat this random process
200 times to obtain the distribution of coefficients and t
statistic values of policy shock. If the coefficient estimates
and tstatistic of HSR are concentrated around 0, it indicates
that policy effect is not random, but it is indeed from HSR.
Table 5: Regression results of instrumental variables.
Variable
Least-cost path-spanning tree
network
Was the railway connected in
1961 Railway passenger traffic in 1998
(1) (2) (3)
HSR second stage 0.394∗∗∗ 0.335∗∗∗ 0.371∗∗∗
(7.59) (3.93) (3.30)
IV reduced regression 0.020∗∗∗ 0.010∗∗∗ 0.008∗∗
(5.42) (2.78) (2.33)
IV first stage for HSR 0.050∗∗∗ 0.030∗∗∗ 0.023∗∗∗
(48.03) (29.21) (21.45)
Control variables Yes Yes Yes Yes Yes Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
N156232 167886 167886 156232 167886 167886 156232 167886 167886
Adj. R-sq 0.119 0.423 0.119 0.402 0.119 0.397
Kleibergen-Paap Fstatistic 5169.985 1890.736 1127.994
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
Table 6: e impact of HSR’s opening on the export: robustness check.
Variable Export volume Per capita export delivery value All cities 2008–2013 Odd years Even years 2000–2016
(1) (2) (3) (4) (5) (6) (7)
HSR 0.056∗∗∗ 0.020∗∗∗ 0.076∗∗∗ 0.044∗∗∗ 0.070∗∗∗ 0.081∗∗∗ 0.092∗∗∗
(3.00) (3.07) (5.06) (2.91) (3.42) (3.37) (5.63)
Control variables Yes Yes Yes Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes Yes Yes No
N167886 164740 211985 109814 84257 83629 201438
Adj. R-sq 0.078 0.744 0.142 0.098 0.166 0.162 0.026
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
Table 7: Robustness check for different geographical distance intervals.
e interval of 14 km e interval of 16 km e interval of 22 km e interval of 22.5 km e interval of 23 km
(1) (2) (3) (4) (5)
[0, 14) 0.083∗∗ [0, 16) 0.087∗∗ [0, 22) 0.117∗∗∗ [0, 22.5) 0.120∗∗∗ [0, 23) 0.114∗∗∗
(2.01) (2.30) (3.71) (3.81) (3.67)
[14, 28) 0.126∗∗∗ [16, 32) 0.068∗[22, 44) 0.057∗∗ [22.5, 45) 0.056∗∗ [23, 46) 0.062∗∗
(3.29) (1.89) (2.18) (2.17) (1.98)
[28, 42) 0.094∗∗ [32, 48) 0.123∗∗∗ [44, 66) 0.009 [45, 67.5) 0.028 [46, 69) −0.003
(2.31) (3.11) (0.25) (0.73) (−0.08)
[42, 56) −0.006 [48, 64) −0.008 [66, 88) 0.081 [67.5, 90) 0.056 [69, 92) 0.122
(−0.13) (−0.16) (1.06) (0.71) (1.52)
[56, 70) −0.000 [64, 80) 0.085 [88, 110) −0.175 [90, 112.5) −0.139 [92, 115) −0.188
(−0.00) (1.06) (−1.29) (−0.98) (−1.32)
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
12 Complexity

Secondly, we advance the time of HSR’s opening by 5 to 8
years. During this period, when HSR was not opened in the
same area, we follow the benchmark model settings and use
data from 2000 to 2007 to perform the model regression
again. If we get the same significant and positive results, it
means that the causality in the benchmark model is not
correct.
e results of the placebo test are shown in Figure 7 and
Table 8. In Figure 7, the coefficients of policy shock obtained
after random processing are distributed between −0.02 and
0.02, and most of the tstatistic values are also between −1.96
and 1.96 (between the dotted lines in Figure 7) and clus-
tering around 0, indicating that the random policy shock of
HSR has almost no impact on exports of agriculture-related
enterprises. In Table 8, at each advanced time, all the co-
efficients of HSR are far less than 0.069, the benchmark
result, and they are not statistically significant. e result of
the benchmark model also has robustness and reliability. e
opening of HSR can significantly promote the export growth
of local agriculture-related enterprises.
6. Further Discussion
In the benchmark regression model, endogenous treatment,
robustness check, and placebo test, we finally concluded that
HSR can bring about growth in agriculture-related enter-
prises’ export. erefore, we will further discuss the
mechanism of HSR on export growth of agriculture-related
enterprises from two aspects: market access and siphonic
effect, combined with distance threshold between enter-
prises and HSR’s stations.
6.1. Mechanism Analysis
6.1.1. Mechanism of Market Access. We learn from
Donaldson to measure the market access caused by HSR,
identifying the mechanism of HSR on export growth of
agriculture-related enterprises, as shown in the following
formula:
MAHSRa�􏽘
b�1
τ−θ
ab GDPb.(18)
Due to the opening of HSR and the improved trans-
portation convenience, the local market access level has been
greatly improved. MAHSR refers to the market access level
brought by the opening of HSR. In formula (18), MAHSRa
represents market access brought by HSR. We set our
benchmark model to deform, and we can get
ln exportict �βmMAHSRct +β2Eict +β3Cct +ξi+ξpt +ξnt +μict.
(19)
In order to explore the impact of market access on the
export of agriculture-related enterprises in different distance
threshold, we divided the distances from enterprises to
HSR’s stations every 15 km. e results are shown in Table 9:
the results in column (1) showed that the construction of
HSR has improved the level of local market access, reduced
the cost of time and space, facilitated local agriculture-
related enterprises to conduct more frequent foreign ex-
changes, strengthened the division of labor and cooperation
between enterprises, and reduced the degree of information
asymmetry, thereby improving the matching efficiency. It
can encourage agriculture-related enterprises to expand
their market search radius, enhance their subjective initia-
tive to enter the international market, and obtain more
export opportunities. In the segmented regression of dif-
ferent geographic distance thresholds, the results in column
(2)–(4) showed that market access brought by HSR to en-
terprises gradually reduced, with the impact on enterprises’
exports within 45 kilometers. HSR has only increased ex-
ports of the agriculture-related enterprises in 0–45 km. is
is consistent with the results in Table 4, which is mainly
because these enterprises are closer to HSR’s station. When
the HSR is opened, they can enjoy the improved market
access brought by HSR in spatial priority and grasp op-
portunities for external communication brought by HSR to
strengthen economic and trade exchanges with outside
market more effectively. When the distance exceeds 45 km,
the agriculture-related enterprises are not efficient in using
HSR, and their communication frequency with outside
regions has little changed. At the same time, higher market
entry barriers also make these enterprises be less willing to
take part in domestic market division of labor cooperation
and international trade, which thus cannot significantly
promote the export.
In order to further identify the impact of market
access caused by HSR on the export, we divided samples
according to regional traffic conditions and enterprise’s
productivity level. Firstly, according to the regional traffic
conditions, the average road traffic level in the top 50% of
the country from 2000 to 2007 is divided into regions with
higher traffic conditions, and the bottom 50% is divided
into regions with lower traffic conditions. Secondly,
according to the level of the enterprise’s productivity, the
top 50% of the annual per capita total industrial output
value is classified as a higher labor productivity of the
enterprise, and the bottom 50% is classified as a lower
labor productivity of the enterprise. e results are shown
in Table 10: in columns (1)-(2), compared to regions with
high-level traffic conditions, HSR has brought bigger
effects of export growth to agriculture-related enterprises
located in regions with low-level traffic conditions. Re-
gions with backward transportation, such as the central
and western provinces of China, have poor economic
foundations and inherently inadequate terms of trade.
HSR provides enterprises in these regions with late-
mover advantages and more opportunities for foreign
exchanges and participation in the industrial division of
labor and cooperation, thus obtaining more international
and domestic market information to promote interna-
tional trade. In columns (3)-(4), compared with low-level
productivity enterprises, high-level productivity enter-
prises can gain greater competitive advantage from the
opening of HSR. According to the heterogeneous-firm
trade theory, high-productivity firms have a higher
tendency of export. ese agriculture-related enterprises
often possess advanced production technology and
Complexity 13

management experience, with a stronger willingness and
tendency to participate in the export. So, they can easily
grasp the policy opportunities of market access brought
by HSR to enter the international market.
6.1.2. Mechanism of Siphon Effect. HSR can improve the
local market access level, reduce communication and in-
formation barriers, and strengthen the understanding of
outside and international market information. However,
HSR will also widen gaps in infrastructure construction
between local regions and surrounding areas without HSR.
Regions with HSR can obtain more communication and
information convenience compared with other places. ey
have more advantages in regional export trade competition,
Table 9: e impact of market access on the export under different geographic distances.
Variable All [0, 15) [15, 30) [30, 45) [45, 60) [60, 75) [75, 90) [90, 105)
(1) (2) (3) (4) (5) (6) (7) (8)
MAHSR 0.003∗∗∗ 0.007∗∗ 0.005∗∗ 0.006∗∗ −0.000 0.002 0.005 −0.014
(2.99) (2.34) (2.26) (2.07) (−0.12) (0.47) (0.63) (−1.02)
Control variables Yes Yes Yes Yes Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes Yes Yes Yes Yes
N166480 34360 36949 30130 24009 12935 8359 5143
Adj. R-sq 0.144 0.153 0.171 0.163 0.193 0.144 0.110 0.170
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
Table 8: Placebo test: time advanced.
Variable Forward_8 years Forward_7 years Forward_6 years Forward_5 years
(1) (2) (3) (4)
HSR −0.026 −0.014 0.014 0.013
(−1.25) (−0.76) (0.61) (0.55)
Control variables Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes
N72764 72764 72764 72764
Adj. R-sq 0.146 0.146 0.146 0.146
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
0
10
20
30
40
50
Distribution
–0.02 –0.01 0 0.01 0.02
Coefficient
0
0.1
0.2
0.3
0.4
Distribution
–4 –2 0 2 4
t statistic
Figure 7: Placebo test: random impact.
Table 10: e impact of market access under different levels of
traffic and productivity conditions.
Variable
Traffic Productivity
High-level Low-level High-level Low-level
(1) (2) (3) (4)
MAHSR 0.003∗∗∗ 0.022∗∗∗ 0.006∗∗∗ 0.000
(2.59) (3.35) (3.42) (0.13)
Control variables Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes
N154713 11767 80485 85995
Adj. R-sq 0.147 0.134 0.136 0.163
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%,
respectively, and the tstatistic value is in parentheses.
14 Complexity

so they may take away export trade opportunities from
surrounding regions. As a result, their agriculture-related
enterprises’ export increases. Meanwhile, exports of enter-
prises in surrounding regions without HSR will relatively
decrease, forming a siphonic effect of export. In this regard,
we verify the siphonic effect mechanism by deform formula
(16):
ln exportift �βnNEARHSRct +β2Eift +β3Cft +ξi+ξpt +ξnt +μift.
(20)
In formula (20), when the region c, which is the closest to
region f, does not open HSR, NEARHSRct �0, and after
opening HSR, NEARHSRct �1. For example, Jiangsu
Nantong did not open HSR, while JiangSu WuXi, the nearest
city to NanTong, opened HSR in 2010. erefore, for JiangSu
NanTong, NEARHSRct �0 before 2010 and NEARHSRct �
1 after 2010. e coefficient βnreflects whether siphonic
effect exists. When βn<0���
s/π
√and it is statistically sig-
nificant, it indicates that HSR in surrounding regions will
have a significant siphonic effect. We also divide the distance
into 7 intervals and perform model regression, respectively.
e results are shown in Table 11; the results in column (1)
indicate that the mechanism of siphonic effect is established.
HSR breaks the balance of regional export behavior, and
agriculture-related enterprises in the region where HSR is
opened can obtain more outside and international market
information, thereby forming a competitive advantage and
relatively obtaining more export opportunity. e results in
columns (2)–(4) show that export of agriculture-related
enterprises in this interval will drop sharply with the opening
of HSR in neighboring regions, resulting in a stronger
siphonic effect. When the distance is longer than 45 kilo-
meters, siphonic effect becomes very small. From the ab-
solute value of the coefficient, we can see that siphonic effect
is greater than market access, which means that, in these two
mechanisms, the siphonic effect of inter-regional competi-
tion triggered by opening HSR is the dominant mechanism
which leads to export growth of agriculture-related enter-
prises. Enterprises with HSR could not only gain greater
trade competitive advantage compared to those regions
where HSR is not open but also seize export opportunities
and orders from similar enterprises in surrounding areas,
thereby strengthening their exports’ growth.
In Table 12, the results in column (1)-(2) demonstrate
that regions with low-level traffic conditions have a bigger
siphon effect after HSR’s opening in adjacent regions. In
recent years, a series of fierce competition about the di-
rection of HSR lines and the distribution of stations have
erupted in China. Regions with HSR have gained economic
and trade competitive advantage over those without HSR.
at is more noticeable in the central and western regions
with poor traffic conditions. In regions with backward
transportation, limited resources are mainly located in re-
gions where HSR is opened, allowing enterprises in such
places to obtain more resource input and trade opportu-
nities. It makes export orders flow from enterprises in re-
gions without HSR to enterprises in regions where HSR was
opened. From perspective of productivity conditions, the
results in columns (3)-(4) show that the siphonic effect
produced by HSR has greater impact on agriculture-related
enterprises with high-level productivity. High-productivity
enterprises have a higher tendency of export and are also
more susceptible to unfavorable market competition
brought by HSR. ese enterprises in regions without HSR
are at a disadvantage in competition with similar enterprises
in regions with HSR, and their exports are relatively
declined.
6.2. Heterogeneity Analysis
6.2.1. Heterogeneity of Enterprises Ownership. As to different
ownership types of agriculture-related enterprises, the im-
pacts of HSR on export may also be different (the main
ownership types of Chinese enterprises are as follows: state-
owned/collective enterprises, Sino-foreign joint/cooperative,
exclusively foreign-owned enterprises, and private enter-
prises). e model estimates are carried out according to
different ownership types, and results are shown in Table 13:
HSR has a significant positive effect on the export of state-
owned/collective, Sino-foreign cooperation/joint, and pri-
vate agriculture-related enterprises, and the most obvious
effect is in state-owned/collective enterprises. However,
exclusively foreign-owned enterprises cannot benefit from
HSR. e main reason is that exclusively foreign-owned
enterprises are more focused on the Chinese market, and the
proportion of domestic sales is higher than that of export, so
export promotion effect brought by HSR is limited. More
importantly, the planning and construction of HSR are
closely related to the local government. Domestic enter-
prises, including state-owned/collective, Sino-foreign co-
operative/joint, and private agriculture-related enterprises,
are more adaptable to local conditions and have actual
advantages in grasping policy changes. ey can accurately
adjust their business and export strategy according to
changes in the external environment.
6.2.2. Heterogeneity of Enterprise Development Stage. We
divide agriculture-related enterprises into three different
development stages, start-up, developing, and maturity (the
start-up are enterprises in the early stages of development.
According to the definition of the Global Entrepreneurship
Observation (GEM) report, they usually refer to enterprises
established within 42 months, that is, within 3.5 years. e
enterprises that have been established for more than 10 years
are relatively mature enterprise, while those in between are
an enterprise that in the developing stage), to discuss the
differences of enterprises’ export in three stages affected by
HSR. Results are shown in Table 14: HSR can effectively
promote export growth of agriculture-related enterprises in
the developing and maturity stage. For the agriculture-re-
lated enterprises of start-ups, HSR cannot bring significant
export growth. e main reason is that, compared with
enterprises in developing and mature stages, most enter-
prises in start-up stages are in danger of survival within 3
years after their establishment, and nearly half of them
survive less than 5 years (the data comes from the “Report on
Complexity 15

the Survival Time of Chinese Domestic Enterprises” in
2013). ey are not yet familiar with the international export
market, and their export behavior is unstable and fragile, so
HSR cannot drive the exports growth of these enterprises.
e business and trade relations of developing and mature
enterprises have gradually stabilized. e opening of HSR
has just provided these enterprises with a larger market
search radius and more potential trade opportunities to
promote export growth.
6.2.3. Regional Heterogeneity of Enterprise. We divide re-
gions into three parts: the east, central, and western parts,
and discuss different impact of HSR on different regions. e
results are presented in Table 15; from the perspective of
policy effects, the coefficients of HSR increase from east to
west, but only in the eastern and western regions are sta-
tistically significant. e eastern region has a geographical
advantage along coast, and HSR has strengthened the
dominant position of local enterprises in industrial chain,
division of labor, cooperation with similar enterprises, and
integrating market information, thereby improving the ef-
ficiency of eastern agriculture-related enterprises in export
trade. In the western region, due to its low level of overall
transportation infrastructure, HSR has gained infrastructure
advantages over surrounding regions, which can greatly
enhance competitive advantage in the international market.
For enterprises in the central region, the impact of HSR on
their exports is not statistically significant. e foremost
reason is that the central region lacks coastal location ad-
vantage compared with the eastern region, and they also lack
of strong trade potential and late-comer advantages
Table 11: e impact of siphonic effect on export under different geographical distances.
Variable All [0, 15) [15, 30) [30, 45) [45, 60) [60, 75) [75, 90) [90, 105)
(1) (2) (3) (4) (5) (6) (7) (8)
NEARHSR −0.074∗∗∗ −0.154∗∗∗ −0.093∗∗ −0.114∗∗∗ −0.012 −0.075 −0.128 0.152
(−4.52) (−3.89) (−2.51) (−2.66) (−0.27) (−1.05) (−1.60) (1.43)
Control variables Yes Yes Yes Yes Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes Yes Yes Yes Yes
N167886 34360 36949 30203 24169 13113 8613 5305
Adj. R-sq 0.143 0.154 0.171 0.163 0.193 0.143 0.112 0.159
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
Table 12: e impact of siphonic effect under different levels of traffic and productivity conditions.
Variable
Traffic Productivity
High-level Low-level High-level Low-level
(1) (2) (3) (4)
NEARHSR −0.069∗∗∗ −0.244∗∗∗ −0.098∗∗∗ −0.033
(−4.05) (−3.26) (−3.88) (−1.46)
Control variables Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes
N155053 12833 81322 86564
Adj. R-sq 0.147 0.129 0.135 0.163
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
Table 13: e impact of HSR’s opening on the export with different enterprises’ ownership.
Variable State-owned/collective Sino-foreign joint/cooperative Exclusively foreign-owned Privately-owned
(1) (2) (3) (4)
HSR 0.214∗∗ 0.075∗∗ 0.028 0.064∗∗
(2.33) (2.11) (0.89) (2.46)
Control variables Yes Yes Yes Yes
Individual FE Yes Yes Yes Yes
Province-year FE Yes Yes Yes Yes
Industry-year FE Yes Yes Yes Yes
N11358 43920 47287 65321
Adj. R-sq 0.137 0.142 0.144 0.179
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%, respectively, and the tstatistic value is in parentheses.
16 Complexity

compared with the western region. With the continuous
upgrading of HSR’s network, many low-end industries
gradually transfer from the eastern region to the central and
western regions. e central region will become an im-
portant gathering place for agriculture-related enterprises
and their export to the international market. At that time,
HSR will play a significant role in promoting the export of
agriculture-related enterprises in the central region.
7. Conclusions and Suggestions
We take the opening of HSR in China as a quasi-natural
experiment, analyze its impact on the export of agriculture-
related enterprises, and then explore the mechanism about
it. On the basis of early research, our research has many
expansions and innovations. In the research on the impact of
transportation infrastructure construction on the economy,
we pay attention to the impact of China’s HSR on trade, and
there are few economists who paid attention to the impact of
China’s HSR; In the discussion of the mechanism of the
impact of transportation infrastructure on exports, early
studies have shown more that the construction of trans-
portation infrastructure can reduce the transportation cost
of enterprises, and enterprises can carry out export activities
at a lower cost, thereby promoting the growth of exports. In
our research, because HSR does not directly transport goods,
we have opened up new mechanisms including market
access and siphon effect to explore the impact of HSR on
exports; in the literature on the China story, the research on
Chinese agriculture is a very important part. Unlike previous
studies, our research combines Chinese agriculture with
China’s HSR and explores the impact of the opening of HSR
on agricultural enterprises.
More specifically, the results of our research show that
HSR can promote the agriculture-related enterprises’ export
growth by 6.9%, and this result is robust because we
get almost consistent results in the regression of changing
different control variables. Meanwhile, the effect of this
policy has continued to increase over time. Furthermore, the
policy effect of HSR is closely related to geographical dis-
tance. HSR has only an effect on the export of enterprises
within 45 km, but when the distance exceeds 45 km, HSR will
no longer have apparent impact on exports. HSR improves
local market access level, strengthens the frequency of
communication, reduces the information barriers to the
outside world, and lowers the cost for obtaining informa-
tion. It can help local agriculture-related enterprises to enter
the international market. At the same time, HSR has
strengthened local infrastructure advantage compared with
regions without HSR, which became a competitive strength
in economic and trade activities. As a result, it has formed a
siphonic effect in export. e policy effect of HSR has
distance threshold for market access and siphonic effect.
When the distance is within 45 km, both the market access
and siphonic effect are established, but when the distance
exceeds 45 km, the HSR has neither market access nor
siphonic effect on export. Compared with the market access,
HSR has a stronger siphon effect on exports for agriculture-
related enterprises. In addition, market access and siphonic
effects are more pronounced in regions with lower primitive
traffic conditions and enterprises with higher productivity.
Finally, the results of heterogeneity analysis show that HSR
has different effects for different types of enterprises; for
agricultural enterprises with different ownership, HSR has a
significant positive impact on the export of state-owned/
collective, Sino-foreign cooperative/joint ventures, and
private agricultural-related enterprises, and the effect is most
obvious in state-owned/collective enterprises. However,
exclusively foreign-owned enterprises cannot benefit from
HSR. For agricultural enterprises at different stages of de-
velopment, HSR can effectively promote the export growth
of agricultural enterprises in the mature stage of develop-
ment. For start-up agriculture-related enterprises, HSR
cannot bring significant export growth. As a populous
country with abundant land resources, China’s agricultural
development is of vital importance. In the process of sus-
tainable agricultural development, HSR also provides a new
opportunity for the increasingly involved agricultural sector,
which has great significance for transformation and
upgrading Chinese agricultural industry and its export-
oriented development path. Armed with this research, we
come up with views and policy recommendations.
Firstly, the results of this paper shows that HSR can
significantly promote the export growth of agriculture-re-
lated enterprises, indicating that the Chinese government
should comprehensively focus on the development oppor-
tunities brought by HSR to the agricultural sector and rural
Table 14: e impact of HSR’s opening on the export with different
development stages.
Variable Start-up Developing Mature
(1) (2) (3)
HSR −0.038 0.093∗∗∗ 0.070∗∗
(−0.64) (4.10) (2.54)
Control variables Yes Yes Yes
Individual FE Yes Yes Yes
Province-year FE Yes Yes Yes
Industry-year FE Yes Yes Yes
N31675 79192 57019
Adj. R-sq 0.231 0.100 0.095
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%,
respectively, and the tstatistic value is in parentheses.
Table 15: e impact of HSR’s opening on the export with different
regions.
Variable Eastern Middle Western
(1) (2) (3)
HSR 0.068∗∗∗ 0.116 0.542∗∗
(3.92) (1.34) (2.12)
Control variables Yes Yes Yes
Individual FE Yes Yes Yes
Province-year FE Yes Yes Yes
Industry-year FE Yes Yes Yes
N154934 9956 2996
Adj. R-sq 0.144 0.159 0.189
Note.∗∗∗,∗∗, and ∗indicate the significance level of 1%, 5%, and 10%,
respectively, and the tstatistic value is in parentheses.
Complexity 17

revitalization, encourage agriculture-related enterprises to
make use of opportunities to integrate agricultural resource
in counties and townships, and vigorously develop export-
oriented agriculture. On this basis, high-quality agriculture-
related enterprises should be encouraged to actively explore
the international market and promote the export of famous
agricultural products to earn income. is will drive Chinese
agriculture to go global.
Secondly, HSR only has an impact on the export of
agriculture-related enterprises within 45 km around HSR
stations. e local government can rationally plan industrial
development based on the location of HSR’s stations and
accelerate the construction of modern industrial parks,
industrial strong towns, and characteristic industrial clusters
near transportation stations. At the same time, local gov-
ernments should guide newly-built enterprises to geo-
graphically get close to transportation stations so as to
improve the convenience of external communication by
means of transportation advantages. e central government
should actively promote the transfer of low-end industries
such as agricultural products processing industries from
developed cities in eastern coastal regions to inland counties
along HSR routes, strengthen the industrial and economic
foundations of counties and townships, and promote the
growth of agricultural exports in these regions.
irdly, there are two main mechanisms for the impact
of HSR on the export of agriculture-related enterprises:
market access and siphonic effect. e market access effect
indicates that we should strengthen the construction of
transportation infrastructure, reduce the local market bar-
riers and restrictions, so as to attract external and inter-
national enterprises to enter the local market for economic
and trade activities, and help more local enterprises enter the
international market. e siphon effect is greater than the
market access effect, indicating that Chinese HSR’s con-
struction is still unbalanced, triggering economic and trade
competition between regions. It is necessary to accelerate the
popularization of HSR networks, rationally plan the layout
of HSR routes, and promote the coverage of HSR in inland
areas, counties, and other backward regions. HSR should be
fully covered in cities and counties where conditions permit.
At the same time, we should vigorously develop cargo
transportation for HSR to reduce the time and cost of cargo
transportation, so as to fundamentally reduce the trade costs
of Chinese agricultural sector and agricultural products.
Finally, HSR has different export effects on different
types of agriculture-related enterprises. It can significantly
promote the export of high-productivity enterprises and
mature enterprises, but it has no impact on the export of
low-productivity and start-up enterprises. e results of
this paper show that we should encourage agriculture-
related enterprises to increase investment in scientific and
technological innovation, continuously improve techno-
logical level and total factor productivity, and transform
from low-end quantity-driven export mode to the high-end
quality-driven export mode. Local governments should
encourage cooperation among different ownership enter-
prises. State-owned enterprises, foreign-funded enter-
prises, and private enterprises should give full play to their
respective advantages and work together to promote export
growth.
Data Availability
e data used to support the findings of this study are
available from corresponding author upon request.
Conflicts of Interest
e authors declare that they do not have any conflicts of
interest.
Acknowledgments
is research was funded by National Natural Science
Foundation of China (Grant nos. 71473282 and 71973046).
References
[1] E. E. Hagen and R. Nurkse, “Problems of capital formation in
underdeveloped countries,” Econometrica, vol. 23, no. 2, 1955.
[2] G. Duranton and M. A. Turner, “Urban growth and trans-
portation,” e Review of Economic Studies, vol. 79, no. 4,
pp. 1407–1440, 2012.
[3] M. Roberts, U. Deichmann, B. Fingleton, and T. Shi, “Eval-
uating China’s road to prosperity: a new economic geography
approach,” Regional Science and Urban Economics, vol. 42,
no. 4, pp. 580–594, 2012.
[4] Y. Wang and B. Wu, “Railways and the local economy: ev-
idence from qingzang railway,” Economic Development and
Cultural Change, vol. 63, no. 3, pp. 551–588, 2015.
[5] A. Banerjee, E. Duflo, and N. Qian, “On the road: access to
transportation infrastructure and economic growth in China,”
Journal of Development Economics, vol. 145, Article ID
102442, 2020.
[6] W. Zou, L. Chen, and J. Xiong, “High-speed railway, market
access, and economic growth,” International Review of Eco-
nomics & Finance, 2019, In press.
[7] M. Diao, “Does growth follow the rail? e potential impact of
high-speed rail on the economic geography of China,”
Transportation Research Part A: Policy and Practice, vol. 113,
pp. 279–290, 2018.
[8] L. Zheng, F. Long, Z. Chang, and J. Ye, “Ghost town or city of
hope? e spatial spillover effects of high-speed railway
stations in China,” Transport Policy, vol. 81, pp. 230–241, 2019.
[9] S. Bougheas, P. O. Demetriades, and E. L. W. Morgenroth,
“Infrastructure, transport costs and trade,” Journal of Inter-
national Economics, vol. 47, no. 1, pp. 169–189, 1999.
[10] J. S. Wilson, C. L. Mann, and T. Otsuki, “Assessing the benefits
of trade facilitation: a global perspective,” e World Econ-
omy, vol. 28, no. 6, pp. 841–871, 2005.
[11] J. Francois and M. Manchin, “Institutions, infrastructure, and
trade,” World Development, vol. 46, pp. 165–175, 2013.
[12] P. Krugman, “Increasing returns and economic geography,”
Journal of Political Economy, vol. 99, no. 3, pp. 483–499, 1991.
[13] K. Behrens, “International integration and regional inequal-
ities: how important is national infrastructure?” e Man-
chester School, vol. 79, no. 5, pp. 952–971, 2011.
[14] G. Duranton, P. M. Morrow, and M. A. Turner, “Roads and
trade: evidence from the US,” e Review of Economic Studies,
vol. 81, no. 2, pp. 681–724, 2014.
18 Complexity

[15] S. Djankov, C. Freund, and C. S. Pham, “Trading on time,” e
Review of Economics and Statistics, vol. 92, no. 1, pp. 166–173,
2006.
[16] B. Hoekman and A. Nicita, “Trade policy, trade costs, and
developing country trade,” World Development, vol. 39,
no. 12, pp. 1–23, 2008.
[17] M. Z. Xu, Riding on the New Silk Road: Quantifying the
Welfare Gains from High-Speed Railways, University of
California, Davis, CA, USA, 2017.
[18] C. Volpe Martincus, J. Carballo, and A. Cusolito, “Roads,
exports and employment: evidence from a developing
country,” Journal of Development Economics, vol. 125,
pp. 21–39, 2017.
[19] D. Niu, W. Sun, and S. Zheng, “Travel costs, trade, and market
segmentation: evidence from China’s high-speed railway,”
Papers in Regional Science, vol. 99, no. 6, pp. 1799–1825, 2020.
[20] N. Limao and A. J. Venables, “Infrastructure, geographical
disadvantage, transport costs, and trade,” e World Bank
Economic Review, vol. 15, no. 3, pp. 451–479, 2001.
[21] H. G. Jacoby and B. Minten, “On measuring the benefits of
lower transport costs,” Journal of Development Economics,
vol. 89, no. 1, pp. 28–38, 2009.
[22] A. K. Cosar and P. D. Fajgelbaum, “Internal geography, in-
ternational trade, and regional specialization,” American
Economic Journal: Microeconomics, vol. 8, no. 1, pp. 24–56,
2016.
[23] D. Donaldson, “Railroads of the raj: estimating the impact of
transportation infrastructure,” e American Economic Re-
view, vol. 108, no. 4-5, pp. 899–934, 2018.
[24] G. Duranton and M. Storper, “Rising trade costs? Agglom-
eration and trade with endogenous transaction costs,” Ca-
nadian Journal of Economics, vol. 41, no. 1, pp. 292–319, 2008.
[25] A. D. Cristea, “Buyer-seller relationships in international
trade: evidence from U.S. states’ exports and business-class
travel,” Journal of International Economics, vol. 84, no. 2,
pp. 207–220, 2011.
[26] A. K. Cosar and B. Demir, “Domestic road infrastructure and
international trade: evidence from Turkey,” Journal of De-
velopment Economics, vol. 81, pp. 232–244, 2016.
[27] B. Faber, “Trade integration, market size, and industrializa-
tion: evidence from China’s national trunk highway system,”
e Review of Economic Studies, vol. 81, no. 3, pp. 1046–1070,
2014.
[28] P. Charnoz, C. Lelarge, and C. Trevien, “Communication
costs and the internal organisation of multi-plant businesses:
evidence from the impact of the French high-speed rail,” e
Economic Journal, vol. 128, no. 610, pp. 949–994, 2018.
[29] A. B. Bernard, A. Moxnes, and Y. U. Saito, “Production
networks, geography, and firm performance,” Journal of
Political Economy, vol. 127, no. 2, pp. 639–688, 2019.
[30] G. M. Grossman and E. Rossi-Hansberg, “Trading tasks: a
simple theory of offshoring,” e American Economic Review,
vol. 98, no. 5, pp. 1978–1997, 2008.
[31] D. Donaldson and R. Hornbeck, “Railroads and American
economic growth: a “market access” approach,” Quarterly
Journal of Economics, vol. 131, no. 2, pp. 799–858, 2016.
[32] M. J. Melitz, “e impact of trade on intra-industry reallo-
cations and aggregate industry productivity,” Econometrica,
vol. 71, no. 6, pp. 1695–1725, 2003.
[33] J. Eaton and S. Kortum, “Technology, geography, and trade,”
Econometrica, vol. 70, no. 5, pp. 1741–1779, 2002.
[34] Y. Lin, “Travel costs and urban specialization patterns: evi-
dence from China’s high speed railway system,” Journal of
Urban Economics, vol. 98, pp. 98–123, 2017.
[35] Y. Qin, ““No county left behind?” e distributional impact of
high-speed rail upgrades in China,” Journal of Economic
Geography, vol. 17, no. 3, pp. 489–520, 2016.
[36] T. Beck, R. Levine, and A. Levkov, “Big bad banks? e
winners and losers from bank deregulation in the United
States,” e Journal of Finance, vol. 65, no. 5, pp. 1637–1667,
2010.
[37] L. Brandt, J. Van Biesebroeck, and Y. Zhang, “Creative ac-
counting or creative destruction? Firm-level productivity
growth in Chinese manufacturing,” Journal of Development
Economics, vol. 97, no. 2, pp. 339–351, 2012.
[38] R. C. Feenstra, Z. Li, and M. Yu, “Exports and credit con-
straints under incomplete information: theory and evidence
from China,” e Review of Economics and Statistics, vol. 96,
no. 4, pp. 729–744, 2014.
[39] S. J. Redding and M. A. Turner, “Transportation costs and the
spatial organization of economic activity,” Handbook of Re-
gional and Urban Economics, vol. 5, pp. 1339–1398, 2015.
[40] W. Jiang, “Have instrumental variables brought us closer to
the truth,” e Review of Corporate Finance Studies, vol. 6,
no. 2, pp. 127–140, 2017.
[41] D. Staiger and J. H. Stock, “Instrumental variables regression
with weak instruments,” Econometrica, vol. 65, no. 3,
pp. 557–568, 1997.
Complexity 19