Railway’s Effects on Economic Development: ...
Railway’s Effects on EconomicDevelopment: A Quantitative Study on Modern Henan (1890s-1930s)
Gao Shuang (高爽)*
Zhengzhou Shengda University of Economics, Business& Management , Zhengzhou, China
Abstract:Different arguments on the relationship between railway construction andeconomic development exist in literature. As the railway construction of Henan inthe period of late Qing Dynasty (1644-1911) and early Republic of China(1912-1949)provides a nearly natural experiment to throw new light on this topic, thispaper creates county-level panel data for this period and conducts a furthertest with a difference-in-difference strategy . While the effect of railway isdemonstrated to be positive in some literature yet negative in others, thispaper identifies some major intergroup differences: railway lines put intoservice after the 1910s and connecting central cities promoted populationgrowth in regions along the railway lines, enhanced their capacity to cope withexternal market shocks, yet they did not improve integration with theintra-provincial market. By contrast, the railway lines put into operation inthe 1930s and connecting hinterland areas not only improved regional marketintegration but contributed to higher living standards as well. The implicationis that the effects of railway are subject to the economic relations of variouslocalities along the railway lines and the economic attributes of varioussectors. thus railways cannot be generally defined to be positive or negative.This conclusion helps explain the disagreements in empirical studies and highlightthe impact of railways on the regional economic structure.
Keywords:late Qing Dynasty and early Republican Period,railway, population density, market integration, welfare level
JEL ClassificationCodes: N75,P23,O18,R11
1.Introduction
Asa modern mode of transport, railway has been closely linked withChina’s socialand economic development since the late 19th century. In the complexhistorical environment, what is railway’s contribution to the economicdevelopment? It needs to be further investigated. Traditional studies onChina’s history of railway construction focused on the issues of foreign debtsand ownership, blaming mismanagement, political turbulence and wars for railwayinefficiency. Despite the recognition of railway’s promotion of agriculturaltrade and export, the relationship between railway and economic development isnot fully proven in existing literature (Xu Dixin, et al., 1990). According to ThomasG. Rawski (2009), new modes oftransport and communications began to transform the pattern of commerce andspurred the output of China’s various economic sectors since the 1890s. Higherproductivity boosted domestic trade and propelled regional and nationaleconomic integration.
However,it is by no means easy to identify the influence of railway. Take theUnited Statesin the 19th century forinstance: the famous “counterfactual” estimate by Robert William Fogel (1962) indicated that thecontribution of railway to GDP was no more than 2%, leading to the conclusionthat railway was not indispensable to the growth of theUSeconomy inthe 19th century. The difficulty of empirical study stems from thequestion of whether railway is an endogenous or exogenous variable in promotingeconomic development, as argued by Albert Fishlow(1965). As a matter of fact, transport infrastructure is both conducive to anddriven by economic development. This presents a challenge in testing thecausality between the two. Yet in the course of economic development, theearly-stage planning and construction of transport routes is highly exogenous.Railway, in particular, is almost synchronous with modern economic development.Early investment on railway is not necessarily based on thorough economic planning.This provides an opportunity to test the relationship between transport andeconomic development. Mindful of this fact, many scholars regard early-stagerailway construction as a natural experiment in their empirical studies, usingsuch methods as difference in differences (hereinafter “DD”) to test the socialand economic influence of railway and thus standardizing Fogel’s counterfactualanalysis (Atack et al., 2009; Michaelet al., 2006; Tang et al., 2013; Li Nan, 2010; Donaldson,2009; Banerjee et al., 2012). Yetdisagreements remain. For instance, in his study onIndiaduring the colonial period, Donaldson (2009) concluded that railwaysignificantly promotes market integration and per capita income growth, whileaccording to another study onChina,Banerjee et al. (2012) is in favor ofFogel’s negative conclusions.
Aside fromthe problem of endogeneity, the influence of railway on economic development isoften unbalanced. Previous literature focused on how railway contributes to grosseconomic growth rather than regional development pattern. Zhang Peigang (1984)noted in his early study that different modes of transport affect the market indifferent ways and that railway has a stronger effect of promoting marketconcentration, but the conclusions of this study are not strictly tested.Meanwhile, discussions on how waterways affect traditional markets arecontroversial as well. In her discussions onChina’s market integration duringthe Qing Dynasty (1644-1911), Mio Kishimoto (2010) posed thequestion of whether such integration is just a linkage of water transport hubsor the formation of a unified regional market covering urban and rural areas.How railway as a modern mode of transport affects economic and market patternsis a crucial aspect of the modernization process. Regretfully, in theliterature on the relationship between railway and economic development,discussions on economic pattern focused on urbanization without responding tothe above question. This is what this paper intends to discuss. In addition,the mechanism and method through which railway exerts its influence on theeconomy also needs to be considered more prudently. Existing literature focuseson whether inter-group differences exist without identifying the technicalattributes of railway transport, locations of routes or associated impacts onthose inter-group differences. As a result, the effects of other relatedfactors may have been attributed to railway and the mechanism and historicalprocess where railway contributed to the economy are not elucidated.
Based onthe above analysis, the difficulties in studying the relationship betweenrailway and economic development are as follows: infrastructure is oftenendogenous to economic development, presenting challenges in dealing with the endogeneity problem; theinfluence of railway and other infrastructures on economic development iscomplex, making it difficult to identify different effects.
Thispaper selects Henan Province in the period of late Qing Dynasty (1644-1911) andearly Republic of China (1912-1949) as the samples for research, because of thestrong exogeneity of railway routes in Henan Province and the relative convenientconditions to identify their impacts, which is shown below in four aspects:
(1)Henan is one of the provinces where theearliest railways were constructed in modernChina. At that time, theconstruction of railway was out of military and political considerations,rather than economic factors. Modern railways in HenanProvince boomed with the invasion of imperialpowers againstChinain the late 19th century that stretched from coastal to inlandprovinces (Yuan Zhongjin, 1993). Under the specific political background atthat time, railway was highly exogenous to economic development.
(2)Neither the Pinghan Railway (between then Peking and Hankou, developed fromLugouqiao-Hankou Railway; the predecessor for today’s Beijing-GuangzhouRailway) nor the Longhai Railway was constructed for Henan’s local development[1]. In the late 19thcentury, the statesman, Zhang Zhidong, put forward a detailed plan for theconstruction of the Pinghan Railway, which was approved by central government.With the primary objective to connect a few major cities, the Pinghan Railwaypassed through Henan Province due to itslocation in the central plain that made it shorter in distance to connect thesecities. A typical example is that the Pinghan Railway did not pass throughKaifeng, provincial capital of Henan Province then, but chose to pass throughZhengzhou for geographical convenience, resulting in the growing political andeconomic clout of Zhengzhou (the current capital of Henan Province) thateventually surpassed Kaifeng.
TheLonghai Railway was originally constructed since October 1889 as a regionalextension of the Pinghan Railway to link Kaifengand Luoyang.Later, the Longhai Railway was further extended into a trunk line connectingChina’s eastern and western regions. It is fair to say thatChina’s modernrailway construction was highly exogenous due to its military and politicalbackground. Railway construction in HenanProvince, which was (and still is) in centralChina, was much more exogenous thanother regions at the time. Figure 1 depicts the distribution of trunk railwaylines passing through Henan Province, which shows the weak relationship betweenrailway construction and the initial distribution of population density (1880),and proves the exogeneity of railway line planning[2].
[1]Daoqing Railway in Henan Provincewas constructed by Fushang Corporation ofBritainfor coal mine developmentand therefore is not entirely exogenous. But this issue does not affect this paper's study, as coaldistribution itself is highly exogenous. Thus, variables like coal are directlyintroduced in the following regression analysis in order to better identify theeffect of railway.
[2] As stated in Cao Shuji’s History of Population in China, Volume 5(Shanghai: Fudan University Press, 2001 edition, page 678- 687), thecatastrophic drought in the second year of Guangxu’s reign spread to HenanProvince but this disaster mainly affected the northern and central parts ofthe province, sparing the south. This is why the population density in thesouth of the province was higher than elsewhere.
(3) After the Grand Canalwas converted into the south-north direction, water transport in Henan Provincedeclined. In the late Qing Dynasty, the Yellow Riverwas seriously silted and its function of water transport diminished. Few otherrivers could support water transport. As described by Sheng Xugong(1931), “the Yellow River is flooded insummers and dries up or becomes frozen in winters. Rising shoals often changeriver course, interrupting water transport.” As a result, railway systems inHenan Province played a more important role in local transport, making itseffect on the economy easier to identify[1].
(4) Due to its location inChina’s central plain, new trunk railway lineswere constructed in Henan Province across variousperiods of time. The Lugouqiao-Hankou line put into operation in 1899, thePeking-Hankou line in 1906 and the Longhai Railway in 1931 formed a basicpattern of crisscrossing vertical and horizontal railways (Local HistoryDrafting Commission of Henan Province, 1987). While the Peking-Hankou RailwayconnectedChina’s centralcities like Beijingand Hankou, the Longhai Railway passed through the hinterland. Such adifference makes it possible to identify the diversity in the relationshipbetween railway and the economy.
As for therelationship between railway construction and economic development in HenanProvince, disagreements exist in literatures, similarly to the national leveldiscussion. In Yuan Zhongjin’s (1993) view, despite control by imperial powersand the impact of wars, railway transport in modern Henan Province still playeda positive role in promoting industrial development, facilitating trade in agricultureand transforming the market structure. Ma Yiping(2010) discussed the positive role of railway in promoting agriculturaldevelopment inChina’scentral region represented by Henan Province. Nevertheless,Wu Yingguang (1992) also mentioned that once railway replaced other modes oftransport, regional markets and large-scale trading activities would alsoagglomerate around the hubs of the new transport lines, while traditionalcenters of business and commerce would decline. In other words, railway did notchange the pattern where traditional markets had been confined to certaincentral hubs. Wu Yingguang (1992) also argued that since much of the land atthese hubs was converted to commercial rather than agriculture, leaving farmerswith little time to engage in grain production, local residents became morevulnerable to famine and economic volatility beyond the control of farmers.This argument implies that railway may not have brought about stable marketintegration.
Hence, thecausality between railway and modern economic development is still academicallydebatable. The contributions of this paper are twofold: first, the highlyexogenous sample of Henan Province in the period of late Qing Dynasty(1644-1911) and early Republic of China (1912-1949) is adopted; second, inaddition to estimating differences among regions with or without railway, howrailway affects the economy is investigated from various dimensions includingpopulation density, market integration and welfare over different periods. Therole of railway was differentiated for China’s central cities and hinterlands.While railways propelled the growth of central cities, its role for thehinterlands is further discussed in this paper.
2. Hypotheses and Methodology
2.1 Hypotheses
Railway construction in Henan Province startedin the late 19th century and then kept extending, bringing differentregions more connected with each other than ever. Theoretically, railway cutscost of transport, improves factor allocation and promotes market integration,thus pushing urbanization and economic development. Specifically, we may arrive atthe following hypotheses:
(1) Railway development facilitates people’s travel and spurs labor flow.If resource allocation were more reasonable and the marginal return to labor were higher along railway lines, more populations wouldhave aggregated in the regions along railway lines, giving rise to regional populationdensity.
(2) Railway significantly cuts the relativedistance and cost of transport between regions, facilitating the transport ofbulk commodities. The grain market in Luohe before PinghanRailway was put into operation was far less prosperous compared with nearbyZhoujiakou and Beiwudu. With the completion of PinghanRailway, freight transport between north and south became unimpeded, resultingin the relocation of grain markets from Zhoujiakou and Beiwudu to Luohe and theexpansion of Luohe’s grain market (Liu Zhisheng, 2002).
If the transport charactersof railway are taken into consideration, railway exerts differentiated impacts onintra-provincial and inter-provincial market integration. Railway isparticularly convenient for the long-distance transport of bulk commodities,particularly transport between major cities. This creates the possibility ofeither coordinating regions along railway lines with central cities orbolstering the integration of hinterlands near railway lines. The two effectsmay not be in balance. If the first effect dominates, cross-regional arbitrage will becomeeasier and price difference across regions will reduce, ironing out pricevolatility and giving rise to inter-provincial markets.
On theother hand, regions along railway lines wouldbecome more sensitive to external volatility and vulnerable to the shocks anduncertainties over a broader geographic range. Their consistency withintra-provincial market was likely to be undermined, giving rise to intra-provincial economicpolarization. As mentioned by Thomas G. Rawski, after the Xihe area of Hebei Provincewas connected with Tianjin and Shanghai by railway, local farmers convertedgrain production to cotton planting, expecting to purchase grain with therevenues of cotton sales to textile factories in both cities. In the event ofprolonged railway transport interruptions due to flood or war, cotton farmersin Xihe area would find themselves in a much worse situation as cotton pricesplunged (Thomas G. Rawski, 1990). Such inter-regionalmarket integration is likely to become a “double-edged sword” described by Mio Kishimoto(2010), arousing vibrant arbitrage, stirring market volatility at the same timeand thus negatively affecting market integration at the local level.
(3) Lower cost of transport will decrease the cost of exporting agriculture productsand increase the gains of agriculture producers,thus leadingto the increase farmers’ wage and other labor. Meanwhile,Adamopoulos (2012) indicated that the reductionof transport cost will promote division of work and specialization and thusraise productivity. With rising efficiency of resource allocation and freemovement of labor across regions and industries, the level of welfare is alsoexpected to improve.Only when population growth along railway linesis accompanied by rising levels of welfare, canurbanization occur as described in the development economics. Otherwise, in HenanProvince in theperiod of late Qing Dynasty (1644-1911) and early Republicof China (1912-1949) when Malthusian threat stillexisted, such population growth was likely to have resulted from logisticalconvenience in surrounding areas of railway lines that attracted displacedrefugees or migrants to railway-connected cities with rising military andpolitical clout and stable social environment.
In this manner, railway can influence moderneconomic development in different mechanisms. Under optimisticassumption, railway will increase population density, bring about marketintegration, improve the level of welfare and spur urbanization and economicgrowth. But the actual effect is subject to different possibilities to be testedrespectively.
2.2 Research Methodology
As demonstrated by existing empirical studies,it is not easy to identify the causality between railway and economicdevelopment. Nevertheless, as mentioned before, if railway construction onlycovers part of the economy and this selected part is exogenous, it can bedeemed as a natural experiment and the causality can be tested by using adifference in difference (DD) model. Bydefining regions with access to railway astreatment group and those without as control group, the basic model canbe specified as follows:
[1] By contrast, water transport was also developed in parallel to railwayconstruction in regions close to rivers and coastlines. Changes in water andmaritime transport must be controlled in order to identify railway’s effect,which increases the difficulty of empirical study.
operation andotherwise 0; county: fixed effect of region; year: fixed effectof period; X: other factors that affect economic development such asnatural disaster. Of course, among these coefficients to be estimated, αis the most concerned in this paper. It is a coefficient of DD term, denotingthe differencebetween the change of railway-crossed region after the railway line is putinto operation and the change of other regions at the same time,with other conditions given. In other words, itmeasures the pure effect of the launch of railway operation, excluding effects of other factors such as differentsituations over time and regions. Despite the strong exogeneity of railwayconstruction in Henan Province to economic development, the correlation betweencross-section factor and railway planning cannot be ruled out. Thus, this paperadopts fixed effect model.
Furthermore, given the non-synchronousconstruction of railway lines across regions, the testis not on a one-time basis and the treatmentgroupcan be further divided accordingto different time spans tocompare the different effects for an early treatment group and a later one, soas to investigate the dynamic effects of railway operation. According to thetime of the launch of railway lines and the distribution of data, this paperdivides sample period into the following three stages: pre-1905, 1910s and1930s. These three phases are selected due to data availability and otherfactors that affect economic development such as war. Most important wars in Henan Province in the period of lateQing Dynasty (1644-1911)and early Republic of China(1912-1949) were the war among warlords in the 1920s and the war on the central plain in the 1930s (Local History Drafting Commission of HenanProvince, 1987). Therefore, the first two sample periods are not affected bymajor wars. As for the sample period of 1930s, we mainly employ the databetween 1933 and 1937, when social stability had been more or less restored. Ofcourse, national situation varied greatly for different sample periods andpolitical stability across individual provinces was uneven as well. The effectsof such differences are brought under control through the fixed effect of timein the econometric model.
The pre-1905 period is the comparison periodwhen the province was not affected by railway operation and in each of thefollowing periods, new railway lines were put into operation in different partsof the province. If these regions are divided into treatment groups by differentperiods of time, e.g. the 1930s group refers to regions where railway lineswere put into service between 1910 and 1930, therewill be different experimental and control groups at different times, which provides various comparisonsto estimate short-term and long-term effects of railway.
Meanwhile, it is also necessary for us toexamine more closely the specific situations of treatment group in light ofactual railway operations;in the regions where railway lines passed through, whether trainstations existed and the number of train stations wouldaffect the role of railway. For instance, Zhengzhou boasted superior advantage of communications as it waslocated at the junction between Pinghan and Longhai railways, which boosted itsdevelopment. Moreover, with different industrial structures, the level ofdependence on railway varied across regions as well. For instance, coaltransport was heavily dependent on railway. Chen Kang (2010) mentioned that thelaunch of Daokou-Qinghua Railway provided an inexpensive mode of coal transportand greatly reduced the cost of transport, cutting the price of coal fromJiaozuo city. Based on these considerations, our model is further specified asfollows:
3.1 Railway Data
Railway is the explanatory variable to the mostconcern of this paper. Various types of transportation historical data haverecorded China’s railway development in detail, which facilitates datacollection. The practice method of this paper is to set the value to be 1 after railway is launched ina county.The railway’s operation of Henan can be dated from 1902, but, since the effect of railway needstime and Pinghan Rail was completed around 1905,we have classifiedall the counties where railway was put intooperation in the first decade after the 20th century as the 1910sgroup, includingcounties whoserailway was constructedbetween 1900 and 1905. Then, the pre-1905 period isjust the comparison period free from railway’s impact. Table 1shows the time opened to traffic of different treatment groups.