The Relationship between Modern Railway Construction and
Financial Market Integration: A Quantitative Study Based on Henan
Province
Shuang Gao
Hubei Business College school of Economics, Wuhan, Hubei, China
Keywords: Modern Railroads, Financial Market Integration, Double Difference Model, Structural Time Series Model.
Abstract: Finance is the bloodline of national economy and the core of modern economy, and the development of
financial market plays a vital role in regional development. The relationship between railroad construction
and financial market integration is less discussed in the literature. This paper takes Henan, which is more
exogenous, as a sample, and constructs financial indicators through recent complete grain price data using the
STSM model; empirically analyzes the difference in the impact of the presence or absence of railroads on
financial markets as well as divides the groups by railroad opening periods to discuss the complex impact of
railroads on financial market integration in different opening periods. It is found that interest rates are lower
in all areas along the railroad, but there are significant group differences. Among the groups by railroad
opening period, the railroads that opened around 1910 and mainly connected to the central cities did not
improve their integration with the provincial financial markets; the railroads that opened in the 1930s and
mainly connected to the hinterland improved the regional financial market integration. This suggests that the
impact of railroads on financial market development is subject to the economic relationships along the route,
the economic attributes of each sector, etc., and that it is not appropriate to generalize. This helps to understand
the divergence in empirical studies, but also highlights the impact of railroads on economic patterns and
urbanization.
1 INTRODUCTION
In the late 19th and early 20th centuries, with the
modern transformation of modern China's economy
and society, a series of new modes of transportation
emerged. The railroad, as one of them, has been
closely associated with the socio-economic changes
since its emergence. Zhang Peigang (1984) mentions
that different means of transportation have different
effects on the market, with railroads tending to
concentrate the market more (Zhang, 1984) In theory,
because railroads reduce transportation costs, they
help optimize factor allocation and promote market
integration. In turn, market consolidation caused by
the reduction of transportation costs is considered to
be an important cause of economic growth in recent
modern times. Therefore, discussing the impact of
railroads on economic development, especially on
market integration, has been a concern of scholars.
Finance is the bloodline of the national economy and
the core of the modern economy, and the
development of financial markets has a crucial role in
regional development. Welfens and Ryan (2011) also
mention the development of financial markets as a
key factor in the economic growth of modern Europe
and America (
Paul, 2011)
.
Due to the lack of complete panel data for recent
interest rate data, testing the relationship between
railroads and financial market integration is a matter
of both data selection and the design of selected
indicators. In the literature, consistency between
prices across locations is usually used to test market
integration. Since only data on grain prices are
relatively systematic before modern times, grain
market integration has the most research. Therefore,
this paper tries to construct a relevant indicator using
grain prices. There have been a number of studies in
this area. As mentioned by McCloskey and Nash
(1984), the storage of grain is actually an investment.
Wheat is stored in October and then in November it
has to pay out the costs of the month, such as storage
costs, depletion costs (spoilage), etc (
McCloskey,
1984)
. If the wheat is sold immediately in November,
102
Gao, S.
The Relationship Between Modern Railway Construction and Financial Market Integration: A Quantitative Study Based on Henan Province.
DOI: 10.5220/0012026500003620
In Proceedings of the 4th International Conference on Economic Management and Model Engineering (ICEMME 2022), pages 102-107
ISBN: 978-989-758-636-1
Copyright
c
2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)
it means that the grain sold and the return on the
money owned (the value of the interest rate) coincide.
If the two do not agree, people will not sell wheat in
November. However, the difference between the two
will converge over time. Therefore, the price of wheat
is equal to the approximate value of interest rates plus
costs such as storage. Based on the above principle,
McCloskey and Nash (1984) used the spread of wheat
prices in medieval England over time to estimate the
interest rate at that time. Assuming that the cost of
storage, etc.is close to a constant, the differences in
seasonal fluctuations in prices across locations are
primarily caused by the interest rate. This method is
used by Mullen Peng (2005, p. 6) to estimate interest
rates using monthly grain prices in the Shandong
states (Peng, 2005). However, the shortcoming of the
above study is that it does not take into account the
exclusion of the non-seasonal component of grain
prices, including the consideration of the unit root, etc.
This paper takes Henan as a sample, one is the
modern The contribution of this paper is to take the
exogenous Henan Province as the sample and to
construct financial indicators using recent complete
data on grain prices; the empirical analysis not only
estimates the difference in the impact of the presence
or absence of railroads on financial markets, but also
divides the groups by railroad opening periods and
discusses the complex impact of railroads on
financial market integration in different opening
periods.
2 HYPOTHESIS AND RESEARCH
METHODOLOGY
2.1 Hypothesis
The railroad construction in Henan, which is located
in the Central Plains, began at the end of the 19th
century with the passage of national railroads, namely
the Ping-Han Railway and the Longhai Railway.
Subsequently, it was continuously expanded and
extended. With the expansion of the railroad network,
different regions became more and more closely
connected to each other. In theory, because railroads
lowered transportation costs, they not only promoted
the flow of goods and labor, but also influenced the
flow of capital. The completion of the Beijing-Han
railroad in 1906 soon stimulated the development of
mining and commercial agriculture along the railroad
lines in the North China Plain, and the demand for
capital would expand. At the same time, the increase
in productivity in the areas along the railroad implied
higher marginal returns to capital, which would lead
to a concentration of capital in these areas. Banerjee,
Duflo, and Qian (2012) argue that this process was
achieved through higher interest rates around the
railroad (Banerjee, 2012). However, significant
economies of scale are usually considered to exist in
finance, and with the increase in the size of financial
markets and the decrease in transaction costs brought
about by capital inflows, capital inflows to areas
along railroads do not require higher interest rates; on
the contrary, interest rates are likely to be lower as a
result.
According to the previous discussion, when the
impact of railroads on economic development is
predominantly positive, d(rail) > 0 corresponds to
d(interest rate) < 0: a decrease in the interest rate.
2.2 Research Methodology
The existing empirical studies show that identifying
the railroad-financial market relationship is not easy
due to the presence of endogeneity. However, as
mentioned earlier, if the construction of railroads
covers only a part of the economy and the selection
of this part is exogenous, it can be considered as a
natural experiment to verify the railroad-financial
market relationship using a double difference (DD)
model. The basic model, with regions with railroad
passage as the experimental group and regions
without railroad passage as the control group, is as
follows.
()
it it t i it it
Interest rate cont rail year county X
(1)
Where the subscript i represents the region and t
represents the period; Interest rate: a variable
measuring the change in interest rates; rail: a dummy
variable with a railroad connection and taking 1 after
opening and 0 otherwise; county: a region fixed effect;
year: a period fixed effect; X: other factors affecting
economic development, such as natural disasters, etc.
Of course, among these coefficients α is the most
concerned in this paper.
Further, combined with the sequence of railroad
construction in time, the experimental group can be
staged and the impact of the experimental group at
different times can be compared, so as to examine the
dynamic impact of railroad construction on the
financial market in different periods. Combining the
opening time of Henan Railway and the availability
of information, this paper divides the sample period
into three periods: before 1905 (pre1905), around
1910 (1910s), and 1930s (1930s). Other factors
affecting economic development, such as war, are
basically excluded from the selection of the sample
period. The wars that occurred in Henan during the
The Relationship Between Modern Railway Construction and Financial Market Integration: A Quantitative Study Based on Henan Province
103
late Qing Dynasty and early Republic of China were
mainly the Warlord Conflict in the 1920s and the
Great War in the Central Plains in 1930s, so the first
two sample periods were not affected by major wars.
As for the sample period of 1930s, the data are mainly
taken from 1933-1937, when the society has basically
restored stability, and the impact of war can be largely
ignored. Of course, the situation was very different
across the country in different periods, and the
stability of the provincial political situation was also
different, and these effects are controlled for in the
econometric model through time fixed effects. In the
three periods, pre1905 is the control period in which
the province was unaffected by the railroad, and each
subsequent period has additional areas open to traffic.
If these areas are divided into different experimental
groups by period - for example, the 1930s opening
group is the areas opened between 1910 and 1930 -
then there will be different experimental and control
groups for different periods, and the short- and long-
term effects of each group can be compared.
At the same time, a more detailed examination of
the specific situation in the experimental group is
necessary in the context of the actual operation of the
railroad. Among them, the presence or absence of
stations and the number of stations in the same
railroad passing area will affect the role of the
railroad. For example, Zhengzhou is located in the
node of the Ping-Han, Longhai Railway, the
transportation location advantage to highlight, the
railroad has become a booster of its development. Not
only that, the industrial structure of different places,
the degree of dependence on the railroad is also
different. (Beijing-Han railroad is located in the main
north-south traffic route, and there are Zhengtai,
Yangluo and Daoqing railroads as branch lines, the
source of goods is wider. The main cargoes
transported by this railroad were "coal and grain",
while coal was the bulk of the transportation in the
northern section, accounting for about half of the
passenger and cargo traffic, and coal produced by
Lincheng coal mine, Jinglong coal mine and coal
mines in Shanxi were directly or indirectly
transported by this railroad. (The main suppliers of
these coals were Beijing, Tianjin and the areas along
the Beijing-Han railroad.) For example, coal
transportation was particularly dependent on the
railroad. Chen Kang mentions that the opening of the
Daoqing Railway provided a cheap way to transport
coal and greatly reduced transportation costs, causing
the price of coal produced in Jiaozuo to drop sharply
in the market. Based on these considerations, the
model is further set as follows.
1980s 1980s
01(1)0itit1itit
1910 1910
( ) + rail station + rail coal
(2)
it p pit p pi t
ps ps
tiitit
Interest rate cont railp railp
year count y X
Where, p represents the group of through traffic;
railp: a dummy variable for the p-period through
traffic, and also introduces its first-order lag term to
examine the long-term impact of the railroad; station:
station density; coal: a dummy variable with coal
companies. The rest is the same as (1).
3 DATA INTRODUCTION
To examine the relationship between railroads and
financial markets based on the model in the previous
section, this paper compiles panel data on the socio-
economy of Henan region from the late 19th century
to the early 20th century. Among them, the sample
period is divided into three periods before 1905
(pre1905), before and after 1910 (1910s), and 1930s
(1930s) as mentioned before. The sample is observed
in counties, and there are 111 counties according to
the administrative division of Henan in the 1930s.
3.1 Railway Data
Railroads are the explanatory variables of most
interest in this paper. The development of railroads is
well documented in various transportation histories,
which facilitates the compilation of this data. The
approach of this paper is to set each county to 1 after
the opening of the railway, so as to obtain the dummy
variable for each opening group. Considering the
time required for the impact of railroad opening, we
put the counties opened to traffic in the first decade
of the 20th century into the 1910s opening group, so
pre1905 is a pure control group not affected by the
railroad.
Due to the limitation of information, the
quantitative indicators such as the mileage of
railroads in each county and the degree of coverage
are not considered. However, what can better reflect
the degree of railroad development in a place is its
station setting status, because the railroad only stops
and carries passengers and goods at the station, and
the setting of railroad stations in the early days was
mainly restricted by exogenous factors such as
location. The Railway Yearbook records in detail the
status of stations of different railroads, including the
time when each station was set up and the time when
the station was converted into a post. Based on this,
this paper collates the number of stations owned by
each county by period and by railroad line. For
example, the number of stations in Zheng County in
ICEMME 2022 - The International Conference on Economic Management and Model Engineering
104
the 1930s actually reached five, three of which were
on the Beijing-Guangzhou line and two on the
Longhai line, which naturally made the convenience
of the railroad in the area much greater than that of
the counties through which a single railroad passed.
However, the number of sites in a certain place may
not necessarily mean that the local passengers and
goods can be more convenient to use the railroad, but
may be simply because the county is too large, the
railroad travels too long miles. Therefore, on the basis
of the number of stations, it is used to divide the area
of each county to calculate the station density of the
county to better measure the convenience of the
railroad.
3.2 Interest Rate Data
In the previous discussion, due to the lack of
complete panel data for interest rate data, this paper
tries to construct a relevant indicator using food
prices.
In terms of data, this paper collates grain prices
from pre1905 to 1930s. Considering the high degree
of commercialization of wheat among the grains in
Henan, wheat prices are selected for the study. The
sources are: pre1905 and 1910s, monthly data of sub-
prefectures from 1895 to 1910 compiled by grain
price transcription files according to the "Grain Price
Table between Daoguang and Xuantong of Qing
Dynasty", with the data of the prefectures to which
they belong instead of the data of each county; 1930s,
monthly prices of sub-prefectures from September
1935 to July 1937 according to the "Henan Monthly
Statistical Report".
In terms of methodology, based on McCloskey
and Nash (1984) and Peng Mullen (2005) studies,
assuming that costs such as storage are close to
constant, the differences in seasonal fluctuations in
prices across locations are mainly caused by interest
rates. However, the shortcoming of the above studies
is that they do not take into account the exclusion of
the non-seasonal component of food prices, including
the consideration of the unit root, etc. Developing a
structural time series model (STSM)
Since grain price data usually contain dynamic
features such as unit roots, it is assumed that for each
period, monthly wheat price data are generated with
the following expression.
2
ti ti ,
2
(1) (1) (1) (1) ,
2
(1) (1) (1) ,
(1)
*
*
(1)
+ 1 111, ( 0, ) (3)
(0, ) (4)
(0, ) (5)
cos sin
sin cos
ti ti ti ti i
ti ti ti ti ti i
ti ti ti ti i
ti
ti c c
ti
ti c c
PiNID
NID
NID
*
/2
,,(1)j(t1)
,
***
1
,,(1)(1)
(6)
cos sin
,(7)
sin cos
ti
ti
s
jj
jti j t i
ti j ti
j
jj
jti j t i jt
wh ere
where P is the price of wheat. Equation (3) means that
it has a trend component μ, a seasonal component γ,
a periodic component φ, and a sum of random
perturbations ε.
The variation of the trend component μ consists
of the horizontal equation (4) and the slope equation
(5).
The periodic component φ is defined by (6),
which has a period of 2π/λc.
The seasonal component γ is defined by (7),
where s denotes the seasonal cycle length.
In the model, the changes in each of the trend,
cycle and seasonal components reflect the normal
dynamic adjustment process of the market, and the
STSM defined in equation (3) precisely allows us to
conveniently extract the seasonal differential in
wheat prices. It is easy to see that the size of the
seasonal spread depends on the variance of the
disturbance term in the seasonal equation of equation
(7), and the standard deviation of the cyclical
fluctuations is calculated from the estimated value of
this variance to measure the level of interest rates.
Table 1 provides a statistical description of each
of these variables by period. As can be seen, there
appears to be no sustained and steady increase in
population density and welfare levels, while market
consolidation has improved over the sample period.
So, what exactly is the role of railroads in this? The
following section will test this through DD analysis.
Table 1: Descriptive Statistics of Variables.
Time Variable
Interest rate
data
Welfare level Drough Flood
Pre1905
AVG 0.029 1.559 0.107 0.039
MAX 0.069 1.874 0.38 0.26
MIN 0.001 1.304 0 0
St
d
0.022 0.154 0.118 0.073
The Relationship Between Modern Railway Construction and Financial Market Integration: A Quantitative Study Based on Henan Province
105
n 106 109 111 111
1910s
AVG 0.032 1.575 0.001 0.391
MAX 0.069 1.909 0.03 1.03
MIN 0.004 1.353 0 0
St
d
0.02 0.162 0.004 0.23
n 111 110 111 111
1930s
AVG 0.026 1.439 0.304 0.042
MAX 0.129 3.281 1.01 0.38
MIN 0 0.919 0 0
St
d
0.031 0.326 0.297 0.083
n 111 104 111 111
328 410 333 333
4 EMPIRICAL ANALYSIS
Table2 gives the estimation results of the causal
relationship between railroads and interest rates
identified using the double difference model.
In model 1, only the standard deviation of cyclical
fluctuations is regressed on the railroad dummy
variable, and the results show a negative coefficient
of the double difference estimator with a significant
level of 10%. That is, interest rates in areas along
railroads are 1.2% lower than those in areas without
railroads. If we look at the access group, the
estimation results of model 2 show that the 1930s
access group has a significantly lower interest rate of
2.2% than its control group, but the 1910s access
group does not have a significant effect on the interest
rate. After adding control variables such as disasters,
in models 3 and 4, the 1910s pass-through group has
a 5.8% lower interest rate than its control group and
is significant at the 10% level, and the 1930s pass-
through group has a 4.8% lower interest rate than its
control group at the 1% level of significance.
Compared to the results in model 2, the coefficients
have increased in absolute value and significant level.
Since there are strong economies of scale in the
financial sector, interest rates decrease in both the
1910s pass-through group, and the 1930s pass-
through group, which rejects the hypothesis of
Banerjee, Duflo, and Qian (2012). Meanwhile, the
1910s station density is negatively but insignificantly
related to the interest rate, then it may be because the
railroad attracts population in this group of areas
mainly through administrative status improvement
and material security improvement, so that the
increase in stations and the convenience of trade do
not play a further impact. As for, the 1930s coal and
interest rates were significantly positive, probably
because the development of the coal industry
increased capital demand, coupled with the remote
location of the mines, which made it difficult for the
development of financial services to follow.
If economies of scale existed, the agglomeration
effect would have been strengthened, so would
interest rates along the railroad have been relatively
and consistently lower? In order to test this dynamic
effect, in model 5, this paper tries again to include the
lag term of 1910s through group. Unfortunately, the
effect of the 1910s opening on the 1930s interest rate
is significantly positive. As explained by Banerjee,
Duflo, and Qian (2012), it is possible that this is due
to the increase in productivity in this group of regions,
which leads to an increase in the return to capital.
However, taking the previous evidence together, it is
more likely that the 1910s opening group (i.e., the
area along the Beijing-Han railroad) did not gain
significant market development and the
agglomeration of the financial sector could hardly
have occurred consistently.
Table 2: Estimated results of railway impact on modern Henan interest rate.
Explained
variable: Interest
rate data
model 1 model 2 model 3 model 4 model 5
Rail
-0.012
*
(
0.006
)
1910s through
g
rou
p
-0.007
(
0.007
)
-0.058
*
(
0.033
)
-0.015
*
(
0.008
)
-0.019
**
(
0.008
)
1930s through
group
-0.022
*
(0.012)
-0.048
***
(0.007)
-0.046
***
(0.007)
-0.040
***
(0.007)
ICEMME 2022 - The International Conference on Economic Management and Model Engineering
106
1910s through
group*
(year=1930s)
0.015
*
(0.009)
Site density *
(
y
ear=1910s
)
-0.007
(
0.005
)
Coal* (year=1930s)
0.025
**
(0.012)
control variable:
Drough
-0.024
**
(0.011)
-0.023
**
(0.011)
-0.018
(0.011)
Flood
0.003
(
0.018
)
0.005
(
0.019
)
0.004
(
0.019
)
Welfare level
-0.023
*
(
0.012
)
-0.025
*
(
0.013
)
-0.028
**
(
0.013
)
Constant term
0.032
***
(0.003)
0.031
***
(0.002)
0.071
***
(0.019)
0.073
***
(0.020)
0.076
***
(0.020)
n 328 328 258 258 258
R
2
0.266 0.272 0.377 0.374 0.381
DW 2.654 2.667 2.776 2.701 2.686
5 CONCLUSION
The relationship between railroads and economic
development has been of considerable interest.
Considering that railroad development is often
endogenous to economic development, this paper
selects a sample with a strong exogenous nature,
modern Henan, to examine the impact of railroads on
financial market integration. Unlike previous
literature, this paper examines the diversification of
the impact of railroads, and the results show that: In
the group divided by the period of railroad opening,
interest rates along the railroad all decrease, the 1930s
opening group improves regional financial market
integration, while the 1910s opening group does not
gain significant market development and financial
agglomeration does not occur consistently.
Interestingly, this paper finds two patterns in the
relationship between railroads and Henan's economic
development: the Beijing-Han line (corresponding to
the 1910s through group), which connects the central
cities of the country, has a limited role in the
development of regional financial markets; in
contrast, the Longhai line (corresponding to the
1930s through group), which mainly connects the
hinterland, had a more robust role in financial
development. Clearly, if urbanization goes hand in
hand with administrative capacity and economic
efficiency, the railroad did not help modern Henan
achieve economic development through this route.
This finding helps us not only to explain the
relationship between railroads and economic
development, but also to reflect on the general path
of economic development. At the same time, the two
modes must be weighed when planning railroad
construction in order to make a reasonable
assessment of their effects.
REFERENCES
Banerjee Abhijit., Esther Duflo and Nancy Qian (2012) On
the Road: Access to Transportation Infrastructure and
Economic Growth in China. Working
Paper17897.cords.
McCloskey and John Nash (1984) Corn at interest: The
Extent and cost of Grain Storage in in Medieval
England American Economic Review,74:1.
Paul J.J. Welfens and Cillian Ryan (2011) Financial Market
Integration and Grown. Springer-Verlag Berlin
Heidelberg.
Peng Mulan (2005) The Construction of the Hinterland:
The State, Society, and Economy of The North China
Mainland (1853-1937). Social Science Literature
Publishing, Beijing.
Zhang, Peigang. (1984) agriculture and industrialization.
Central China Institute of Technology Press. Wuhan.
The Relationship Between Modern Railway Construction and Financial Market Integration: A Quantitative Study Based on Henan Province
107
