Trans­porta­tion In­fra­struc­ture and Pro­duc­tiv­ity Growth: Ef­fects of Rail­way Speed-Up on Firm’s TFP in China

ShiZhenkai(施震凯),ShaoJun(邵军)andPuZhengn­ing(浦正宁)

China Economist - - Contents - By the def­i­ni­tion of the In­ter­na­tional Union of Rail­ways (UIC), high-speed rail­way refers to new lines de­signed for speeds above 250 km/h and in some cases, up­graded ex­ist­ing lines for speeds up to 200 or even 220 km/h.

Shi Zhenkai (施震凯), Shao Jun (邵军) and Pu Zhengn­ing (浦正宁) School of Eco­nom­ics and Man­age­ment, South­east Univer­sity, Nan­jing, China Ab­stract:

Im­prove­ment of trans­porta­tion in­fra­struc­ture qual­ity will lead to more suf­fi­cient mar­ket com­pe­ti­tion and pro­mote the flow of re­sources with greater ef­fi­ciency. This paper con­sid­ers China’s rail­way speed-up in 2007 as a quasi-nat­u­ral ex­per­i­ment on China’s trans­porta­tion in­fra­struc­ture qual­ity im­prove­ment. With the ini­tial op­er­a­tion of elec­tric mul­ti­ple units (EMUs) as the ba­sis of group­ing, this re­search ex­am­ines the ef­fect of rail­way speed-up on cor­po­rate to­tal fac­tor pro­duc­tiv­ity (TFP) growth by the dif­fer­en­cein-dif­fer­ences (DID) method. Over­ally, the re­sults re­veal pos­i­tive ef­fects both on firms’ tech­no­log­i­cal change and ef­fi­ciency im­prove­ment, which lead to the in­crease of TFP. Based on sub­sam­ples di­vided by dif­fer­ent re­gions and types of en­ter­prises, fur­ther anal­y­sis in­di­cates that the pro­duc­tiv­ity of ex­porter, non-state and coastal firms has been mostly af­fected by the rail­way speed-up. These con­clu­sions are ver­i­fied by a placebo test. Be­sides, firms within “one-hour eco­nomic cir­cle” have been shown more sen­si­tive to the ef­fect of rail­way speed in­crease.

Key­words:

rail­way speed-up, to­tal fac­tor pro­duc­tiv­ity (TFP), data en­vel­op­ment method, dif­fer­ence-in-dif­fer­ences method

JEL Clas­si­fi­ca­tion Codes: D24; O38

DOI:1 0.19602/j .chi­nae­conomist.2018.11.0219

1. In­tro­duc­tion

World eco­nomic his­tory since the In­dus­trial Revo­lu­tion clearly shows that trans­porta­tion in­fra­struc­ture im­prove­ment is es­sen­tial for eco­nomic growth. For China, a large coun­try with com­plex ge­o­graph­i­cal char­ac­ter­is­tics, rail­way in­fra­struc­ture is of great sig­nif­i­cance to its eco­nomic and so­cial de­vel­op­ment. Since re­form and open­ing up, China has made re­mark­able progress in its rail­way con­struc­tion. The to­tal length of rail­way op­er­a­tion in China in­creased from 53,000 kilo­me­ters in 1979 to 124,000 kilo­me­ters in 2016, re­sult­ing in a rail­way net­work with pro­vin­cial cities as nodes con­nect­ing other ma­jor cities na­tion­wide. How­ever, China’s per capita rail­way net­work den­sity is far below the level of de­vel­oped coun­tries. Given the in­suf­fi­ciency of trans­porta­tion ca­pac­ity, in­creas­ing train op­er­a­tion speed be­came an im­por­tant so­lu­tion. The for­mer Min­istry of Rail­way con­ducted six rail­way speed-ups from 1997 to 2007. As a re­sult, the av­er­age train speed in China grad­u­ally in­creased from 48 km/h to

70 km/h. In 2007, China put into ser­vice the elec­tric mul­ti­ple units (EMU) op­er­at­ing at above 200 km/ h. Over a short pe­riod of time, China con­structed more than 6,000 kilo­me­ters of high-speed rail­way , un­veil­ing a new era of high-speed trains.

Over the years, the role of trans­porta­tion in­fra­struc­ture in eco­nomic growth did not re­ceive suf­fi­cient at­ten­tion in eco­nomic re­search. With re­mark­able im­prove­ment in China’s trans­porta­tion in­fra­struc­ture, rel­e­vant lit­er­a­ture stud­ies are in­creas­ing. The­o­ret­i­cally, trans­porta­tion in­fra­struc­ture im­prove­ment will ac­cel­er­ate the stream of peo­ple, pro­mote fixed as­set in­vest­ment, in­crease cap­i­tal stocks, en­hance cap­i­tal deep­en­ing, and boost eco­nomic growth rate through spillover ef­fects. How­ever, all these are short-term ef­fects. To­tal fac­tor pro­duc­tiv­ity (TFP) is the fun­da­men­tal driver of long-term eco­nomic growth. Un­der nat­u­ral re­source lim­i­ta­tions and en­vi­ron­men­tal con­straints, it is un­sus­tain­able for eco­nomic growth to solely rely on pro­duc­tion fac­tor in­put; thus, eco­nomic growth in­creas­ingly needs to be driven by TFP. Some stud­ies have ex­am­ined the pro­duc­tiv­ity ef­fects of trans­porta­tion in­fra­struc­ture. For in­stance, Aschauer (1989) in­ves­ti­gated the TFP ef­fects of high­ways in the United States, and Farhadi (2015) un­cov­ered the TFP ef­fects of trans­porta­tion in­fra­struc­ture in 18 OECD coun­tries. Both stud­ies re­veal the pos­i­tive role of trans­porta­tion in­fra­struc­ture. With grow­ing avail­abil­ity of mi­cro­data, an in­creas­ing num­ber of stud­ies have been car­ried out to in­ves­ti­gate the cor­po­rate TFP ef­fects of trans­porta­tion in­fra­struc­ture. How­ever, most stud­ies of this kind are fo­cused on high­ways. For in­stance, Gib­bons et al. (2012) dis­cov­ered that an im­prove­ment in high­way qual­ity may have an im­por­tant in­flu­ence on cor­po­rate em­ploy­ees, la­bor pro­duc­tiv­ity and other as­pects. Li and Li (2013) dis­cov­ered that an in­crease in high­way in­vest­ment will save firms’ ware­hous­ing cost and in­crease pro­duc­tiv­ity. Holl (2016) dis­cov­ered that high­ways pro­mote cor­po­rate TFP growth by in­creas­ing the re­gional den­sity and clus­ter­ing ef­fect. How­ever, very few stud­ies, es­pe­cially stud­ies on firms at a mi­cro­scopic level, have been car­ried out to un­ravel the TFP ef­fects of rail­way in­fra­struc­ture. By ad­dress­ing this topic, this paper at­tempts to fill this in­for­ma­tion gap in ex­ist­ing re­search lit­er­a­ture.

In or­der to an­a­lyze the TFP ef­fects of rail­way in­fra­struc­ture im­prove­ment on firms, we should mea­sure the im­prove­ment of trans­porta­tion in­fra­struc­ture qual­ity, which is mea­sured in ex­ist­ing stud­ies by such cri­te­ria as in­creased in­vest­ment, op­er­a­tion length, trans­porta­tion ca­pac­ity and re­duced time of op­er­a­tion. For in­stance, Deng et al. (2014) ex­am­ines the eco­nomic growth ef­fects of trans­porta­tion in­fra­struc­ture based on in­vest­ment and mileage data. Huang and Xu (2012) dis­cov­ered that re­duc­tions in trans­porta­tion cost, such as ex­emp­tion of high­way toll fees, are con­ducive to ex­ports in in­te­rior re­gions. Based on China’s trans­porta­tion in­fra­struc­ture and in­vest­ment stock data, Shi and Huang (2014) in­ves­ti­gated whether China’s trans­porta­tion in­fra­struc­ture is in over­sup­ply rel­a­tive to eco­nomic de­vel­op­ment. Many stud­ies re­gard China’s rail­way speed-up as a quasi-nat­u­ral ex­per­i­ment for the anal­y­sis of mul­ti­fac­eted ef­fects of rail­way in­fra­struc­ture im­prove­ment. For in­stance, Zhou and Yu (2013) dis­cov­ered that China’s rail­way speed-up has a more sig­nif­i­cant pos­i­tive ef­fect on sec­ondary in­dus­try than on ter­tiary in­dus­try. Song and Li (2015) found that China’s rail­way speed-up will sig­nif­i­cantly pro­mote pop­u­la­tion growth in cities along the rail and high­way routes in the long run. Ref­er­enc­ing the above meth­ods, this paper iden­ti­fies China’s rail­way speed-up as a quasi-nat­u­ral ex­per­i­ment of trans­porta­tion in­fra­struc­ture im­prove­ment to ex­am­ine its TFP ef­fects.

Specif­i­cally, this paper cre­ates a treated group and a con­trol group based on whether a city has ac­cess to EMU trains. Based on that, the study is car­ried out us­ing dif­fer­ence-in-dif­fer­ences method. Treated group and con­trol group sam­ples are dif­fer­en­ti­ated by whether cities have ac­cess to EMU train sta­tions rather than whether cities are along rail­way route, since not all cities along rail­way route have ac­cess to EMU trains. The most im­por­tant mean­ing of rail­way speed-up is the op­er­at­ing route. How­ever,

dif­fer­ent from high­ways for which en­trance and exit ramps can be cre­ated where nec­es­sary, rail­way sta­tions may only be con­structed at op­ti­mal sites due to cost con­sid­er­a­tions. Whether a city has a rail­way sta­tion de­ter­mines the city’s ac­cess to a rail­way trans­porta­tion net­work. What mat­ters is not only whether a city is lo­cated along a rail­way route or the rail­way length within its ju­ris­dic­tion. Rather, the key is whether it has a rail­way sta­tion with ac­cess to rail­way net­work. Cities with rail­way sta­tions pro­vide more op­tions for the flow of peo­ple, lo­gis­tics and in­for­ma­tion com­pared with oth­ers with­out rail­way sta­tions.

2. The­o­ret­i­cal Mech­a­nism

Im­prove­ment of firm’s TFP stems from tech­nol­ogy progress and man­age­rial in­no­va­tion. The role of tech­nol­ogy progress is par­tic­u­larly im­por­tant. New tech­nol­ogy is of­ten cre­ated in one place, and spills over to other places (Liu et al., 2010). How­ever, tech­nol­ogy dis­sem­i­na­tion is both a tem­po­ral and spa­tial process. The longer it takes to dis­sem­i­nate, the more in­for­ma­tion will be dis­torted, i.e. longer dis­tance di­min­ishes the tech­nol­ogy spillover ef­fect (Fu, 2009). Spa­tial dis­tance in the mod­ern sense is not en­tirely con­fined to phys­i­cal dis­tance. It is also sub­ject to tem­po­ral fac­tor. This paper in­tro­duces real dis­tance de­flated by tem­po­ral fac­tors into Keller’s (2002) model to de­ter­mine pro­duc­tiv­ity’s re­la­tion­ship with real dis­tance and ex­plains the in­flu­ence of rail­way speed-up. As­sum­ing that Place A and Place B mu­tu­ally in­flu­ence each other and are sym­met­ric, their ge­o­graph­i­cal dis­tance is D, and rail­way speed-up will shorten train op­er­a­tion time from t0 to tr. We use t=tr/t0 to mea­sure the im­pact of time on ge­o­graph­i­cal dis­tance. Smaller t means shorter real dis­tance be­tween the two places. Hence, the real dis­tance of trans­porta­tion de­flated by t can be ex­pressed as Dr=tD. As­sum­ing that the out­put Y of firms in Place A meets:

Where, A is con­stant term, K is cap­i­tal, and α is out­put elas­tic­ity, and 0< α< 1. m and n are the quan­ti­ties of in­ter­me­di­ate in­puts pro­vided by mo­nop­o­lis­tic man­u­fac­tur­ers in Place A and Place B. c and c* are the types of in­put. N and N* are cor­re­spond­ing in­put port­fo­lios. As­sum­ing that the prices p and p* of in­ter­me­di­ate in­puts in both places are

Man­u­fac­tur­ers in both places re­spec­tively use la­bor L and L* as the only in­put fac­tor. w and w* are the wage lev­els of both places. Cost of trans­porta­tion for firms in Place A to use lo­cal prod­ucts is 0. How­ever, the ice­berg trans­port cost of trans­porta­tion for use of non­lo­cal prod­ucts is . Thus, equa­tions and hold true un­der the op­ti­mal con­di­tion. When general equi­lib­rium is reached, we may fur­ther de­duce the fol­low­ing re­la­tional ex­pres­sion be­tween m and n from p=p* and w=w*.

Namely, the elas­tic­ity of sub­sti­tu­tion be­tween lo­cal in­put prod­uct m and non­lo­cal in­put prod­uct n em­ployed by firms in Place A is 1/(1- α). Given the use of non-sub­sti­tutable la­bor L and L* as in­put fac­tor in the two places and the in­ter­me­di­ate in­put port­fo­lios of N and N* re­spec­tively, we may fur­ther ar­rive at:

3. Model and Data

After the first rail­way speed- up ad­min­is­tered by the for­mer Min­istry of Rail­way for Bei­jingGuangzhou Rail­way and Beijing-Shang­hai Rail­way in April 1997, six rail­way speed-ups were car­ried out prior to 2007. But the first five rail­way speed-ups were car­ried out only for tra­di­tional lo­co­mo­tives with very lim­ited speed in­creases. In com­par­i­son, the rail­way speed-up in 2007 rep­re­sents a ma­jor mile­stone that un­veiled EMU trains that op­er­ate at over 200 km/h. This paper re­gards China’s rail­way speed-up of 2007 as a quasi-nat­u­ral ex­per­i­ment and ex­am­ines the cor­po­rate TFP ef­fects of rail­way in­fra­struc­ture qual­ity im­prove­ment us­ing the dif­fer­ence-in-dif­fer­ences method.

In con­duct­ing dif­fer­ence-in-dif­fer­ences method, the treated group and con­trol group should be set first. Ac­cord­ing to the train sched­ule of April 2007, we iden­tify 58 sta­tions along EMU routes lo­cated in 49 cities in­volv­ing 10 prov­inces, 3 mu­nic­i­pal­i­ties and 1 spe­cial eco­nomic zone. Mu­nic­i­pal­i­ties and spe­cial eco­nomic zones are more de­vel­oped and highly het­ero­ge­neous com­pared with other cities. Thus, it is dif­fi­cult to cre­ate a con­trol group com­pat­i­ble with them. For this rea­son, mu­nic­i­pal­i­ties and spe­cial eco­nomic zones are ex­cluded from the ex­am­ples, so there are 45 cities left.. Then, firms lo­cated in these cities are clas­si­fied into treated group, and other cities in these 10 prov­inces are taken into the con­trol group. We iden­tify the ge­o­graph­i­cal dis­tri­bu­tion of the treated group and con­trol group in Ta­ble 1. Specif­i­cally, the coastal re­gion in­cludes 5 prov­inces, i.e. He­bei, Shan­dong, Jiangsu, Zhe­jiang and Guang­dong. The re­main­ing 5 prov­inces are Jiangxi, He­nan, Hubei, Hu­nan and Shaanxi, which are in­land re­gions.

After set­ing treated group and con­trol group, this paper cre­ates the fol­low­ing re­gres­sion model based on the dif­fer­ence-in-dif­fer­ences method:

Where, Z is a de­pen­dent vari­able and refers to TFP-re­lated vari­ables ob­tained by DEA method, in­clud­ing Malmquist In­dex ( mqst), tech­nol­ogy change in­dex ( mtc) and ef­fi­ciency change in­dex ( mtec). CRH×pe­riod is the most im­por­tant in­de­pen­dent vari­able in the model, and its value is the prod­uct

be­tween CRH and pe­riod. CRH is a group-spe­cific dummy vari­able that de­scribes dif­fer­ences be­tween treated group and con­trol group. If a firm is lo­cated in a re­gion in­volved in the rail­way speed-up, it is clas­si­fied as treated group with the value of 1; oth­er­wise, the value is 0. pe­riod is time dummy vari­able that dis­tin­guishes the dif­fer­ences be­fore and after rail­way speed-up, and its value is 1 dur­ing 2007-2009 and 0 for other years. h is the fixed ef­fect of in­dus­try, u is the fixed ef­fect of re­gion, t is the fixed ef­fect of time, cons is con­stant term, and ε is dis­tur­bance term. Be­sides. X mainly in­cludes the fol­low­ing con­trol vari­ables: kl is cap­i­tal den­sity mea­sured by the ra­tio be­tween cor­po­rate cap­i­tal and em­ploy­ees; sub is a dummy vari­able that denotes whether a firm en­joys govern­ment sub­sidy, and its value is 1 if it does; oth­er­wise, it is 0. Firm size (scale) is a dummy vari­able whose value is 1 for large and medium-sized firms, and 0 for small firms. The city de­vel­op­ment ( citygdp) is mea­sured by the GDP of the cities. Data em­ployed in this paper are pri­mar­ily from the China In­dus­trial En­ter­prises Data­base 2001-2009 and the China City Sta­tis­ti­cal Year­books.

4. Re­gres­sion Re­sult and Anal­y­sis

4.1 Ba­sic Re­sults and Anal­y­sis

An im­por­tant pre­con­di­tion for test­ing pol­icy ef­fects us­ing the dif­fer­ence-in-dif­fer­ences method is to meet the “par­al­lel trend hy­poth­e­sis,” i.e. there is no sys­tem­atic dif­fer­ence be­tween a treated group and con­trol group. Be­fore the oc­cur­rence of an event, both of the groups share con­sis­tent de­vel­op­ment char­ac­ter­is­tics and trends. Oth­er­wise, the re­sult of dif­fer­ence-in-dif­fer­ences method is likely to be un­be­liev­able. Ref­er­enc­ing the meth­ods of Moser and Voena (2012) and Tanaka (2015), this paper draws a time se­ries plot of key vari­ables and es­ti­mates par­al­lel trend to as­sess whether treated group and con­trol group sam­ples share a con­sis­tent trend be­fore rail­way speed-up. First, we drew a time se­ries plot (Fig­ure 1) for the mean val­ues of de­pen­dent vari­ables of treated group and con­trol group. Through ob­ser­va­tion of Fig­ure 1, we can dis­cover that, de­spite the slight dif­fer­ences in the time se­ries curves of vari­ables prior to the rail­way speed-up, the in­creases and de­creases of the treated group and con­trol group share a con­sis­tent trend and demon­strate rel­a­tively ro­bust par­al­lel char­ac­ter­is­tics.

Fig­ure 1 pro­vides an ini­tial as­sess­ment that treated groups and con­trol groups share a sim­i­lar de­vel­op­ment trend prior to rail­way speed-up im­ple­men­ta­tion. How­ever, par­al­lel trend as­sump­tions re­quire a more pre­cise test. Re­fer to Moser and Voena (2012), we fur­ther in­tro­duce a cross-mul­ti­ply­ing term be­tween the group-spe­cific dummy vari­able and the time trend and re­port the re­sults in Ta­ble 2. It can be found that after con­trol­ling for in­dus­try ef­fect, time ef­fect, re­gional ef­fect and con­trol vari­ables, de­spite the pos­i­tive and nega­tive val­ues of the re­gres­sion co­ef­fi­cients of “treated group × year” term, the dif­fer­ences in such val­ues are in­signif­i­cant at the 10% sta­tis­ti­cal level and can­not re­ject the null hy­poth­e­sis that a treated group and con­trol group will share a con­sis­tent de­vel­op­ment trend prior to the rail­way speed-up. Thus, the par­al­lel trend hy­poth­e­sis test is passed. This re­sults above mean that the

ex­per­i­men­tal group­ing in this paper is ap­pro­pri­ate, and that it is rea­son­able to use the dif­fer­ence-in­dif­fer­ences method to de­ter­mine the ef­fects of the rail­way speed-up on firm’s TFP.

Ta­ble 3 pro­vides the re­gres­sion re­sults based on the dif­fer­ence-in-dif­fer­ences method. To make a com­par­a­tive anal­y­sis, this paper si­mul­ta­ne­ously pro­vides two re­sults, one for core in­de­pen­dent vari­ables and an­other for in­clu­sion of con­trol vari­ables. Ap­par­ently, there is no sig­nif­i­cant change in the co­ef­fi­cients of CRH × pe­riod. For the two re­gres­sion equa­tions of the Malmquist in­dex ( mqst), the co­ef­fi­cients of CRH × pe­riod are all sig­nif­i­cantly pos­i­tive. This re­sult shows that the rail­way speedup of 2007 had a sig­nif­i­cantly pos­i­tive ef­fect on cor­po­rate TFP growth. Col­umns 4 and 5 of Ta­ble 3 are the re­gres­sion re­sults of the tech­nol­ogy change in­dex ( mtc) as a de­pen­dent vari­able. More­over, the co­ef­fi­cients of CRH × pe­riod are sig­nif­i­cantly pos­i­tive, i.e. the rail­way speed-up had a pos­i­tive ef­fect on the tech­nol­ogy of firms in rel­e­vant re­gions. The last two col­umns of Ta­ble 3 are the es­ti­ma­tion re­sults for the ef­fi­ciency change in­dex ( mtec) and the co­ef­fi­cients of CRH×pe­riod are all in­signif­i­cantly pos­i­tive, mean­ing the rail­way speed- up has an in­signif­i­cantly pos­i­tive ef­fect on firm ef­fi­ciency im­prove­ment. While the rail­way speed-up sig­nif­i­cantly boosted firms’ tech­nol­ogy progress in rel­e­vant re­gions, its ef­fi­ciency im­prove­ment ef­fect was not sig­nif­i­cant. A pos­si­ble rea­son is the par­tially nega­tive ef­fi­ciency ef­fect of the rail­way speed-up. Over­all, de­spite the in­signif­i­cant ef­fi­ciency im­prove­ment

due to un­cer­tain­ties of rail­way speed-up, the tech­nol­ogy of firms ben­e­fited from the in­creased flows of pas­sen­gers, lo­gis­tics and in­for­ma­tion, and the pos­i­tive ef­fects of rail­way speed-up should be rec­og­nized.

4.2 Het­ero­ge­neous Ef­fects

(1) We ex­am­ine the ef­fects of rail­way speed-up on SOE and non-SOE’s TFP, and the re­sults are shown in Ta­ble 4. It can be found that the rail­way speed-up had a sig­nif­i­cant pos­i­tive ef­fect on the tech­nol­ogy and ef­fi­ciency of non-SOEs; the re­gres­sion co­ef­fi­cients of CRH × pe­riod are 0.0093 and 0.0023 re­spec­tively, which pro­mote cor­po­rate TFP growth un­der their cou­pled ef­fect. While the rail­way speed-up boosted SOE ’s tech­nol­ogy, the TFP is in­signif­i­cant due to sig­nif­i­cant ef­fi­ciency re­duc­tion. To some ex­tent, the ef­fect is nega­tive. As men­tioned be­fore, rail­way speed-up may sig­nif­i­cantly in­crease

mar­ket com­pe­ti­tion in rel­e­vant re­gions. Due to the agent-prin­ci­pal re­la­tion­ship of SOEs and other prob­lems like soft bud­getary con­straints, it takes a long time for SOEs to adapt to a new com­pet­i­tive en­vi­ron­ment. De­spite tech­nol­ogy im­prove­ment after rail­way speed-up, SOE’s ef­fi­ciency re­duced sharply due to nega­tive ef­fects. Un­der the goal of profit max­i­miza­tion, pri­vate firms are more sen­si­tive to changes in mar­ket struc­ture and ex­ter­nal­i­ties aris­ing from rail­way speed-up, which is con­ducive to their TFP growth.

(2) We clas­sify sam­ples into coastal and in­land re­gions for re­gres­sion and present re­gres­sion re­sults in Ta­ble 5. Judg­ing by the re­sult of re­gres­sion of mtc, the co­ef­fi­cients of the CRH × pe­riod are all sig­nif­i­cantly pos­i­tive for both coastal or in­land re­gions. In the re­gres­sion of mtec, the co­ef­fi­cients of CRH × pe­riod are all in­signif­i­cant. How­ever, the val­ues of the co­ef­fi­cients are pos­i­tive for coastal

re­gions and nega­tive for in­land re­gions, and the ab­so­lute val­ues of the lat­ter are more sig­nif­i­cant. In sum­mary, the cou­pled ef­fect of mtc and mtec in­flu­ences change in mqst. By the re­sults of the mqst model, we may find that the TFP growth of firms in coastal cities cov­ered by the rail­way speed-up cam­paign re­sponded pos­i­tively to the rail­way speed-up. How­ever, the ef­fect is lim­ited for firms lo­cated in in­land re­gions.

(3) Ta­ble 6 lists the re­gres­sion re­sults for firms whose tar­get mar­ket is the do­mes­tic mar­ket and firms whose tar­get mar­ket is the ex­port mar­ket. The ab­so­lute value of the CRH × pe­riod in the mtc model of ex­port firms is greater than the non-ex­port firms and is more sig­nif­i­cant. Rail­way speedup has dif­fer­en­ti­ated ef­fects on the ef­fi­ciency im­prove­ment of non-ex­port firms and ex­port firms. The dif­fer­ences may have to do with the trans­porta­tion modes used by the two types of firms. At present,

marine trans­porta­tion re­mains the pri­mary mode of trans­porta­tion for ex­ports. On key in­ter­na­tional marine trans­porta­tion routes, most goods suitable for con­tainer trans­porta­tion are trans­ported us­ing con­tain­ers. When in­land man­u­fac­tur­ers ex­port goods, they will opt for full con­tainer loads (FCLs) or less con­tainer loads (LCLs) to ship con­tain­ers di­rectly to ports. Trans­porta­tion ca­pac­ity re­leased by rail­way speed-up is fa­vor­able to con­tainer trans­porta­tion and helps ex­port com­pa­nies to ac­cel­er­ate in­ven­tory turnover and in­crease ef­fi­ciency. In ad­di­tion, ex­ist­ing stud­ies dis­cover that an in­crease of com­mod­ity trans­porta­tion time by one day is equiv­a­lent to an in­crease of com­mod­ity price by 0.6%–2.1%. In ad­di­tion, long trans­porta­tion de­lays will sig­nif­i­cantly re­duce the suc­cess rate of ex­ports (Hum­mels and Schaur, 2013). Rail­way speed-up will re­duce the cost of trade for ex­port firms and is fa­vor­able to their ef­fi­ciency im­prove­ment. For firms whose tar­get mar­ket is the do­mes­tic mar­ket, long-haul trucks us­ing

high­ways are more con­ve­nient than rail­way trans­porta­tion. There­fore, rail­way speed-up has a rel­a­tively lim­ited ef­fect for non-ex­port firms with in­signif­i­cant im­pact on their ef­fi­ciency im­prove­ment

4.3 Ro­bust­ness Test

(1) Placebo test. Some other poli­cies or sto­chas­tic fac­tors may also af­fect firm’s TFP. If such an im­pact is not cor­re­lated with rail­way speed-up, CRH×pe­riod should re­main sig­nif­i­cant in other years. This im­plies that the con­tri­bu­tion of CRH×pe­riod to cor­po­rate TFP should be at­trib­ut­able to fac­tors other than rail­way speed-up. So we con­duct a placebo test by al­ter­ing pol­icy im­ple­men­ta­tion time to iden­tify the doubt above. With­out chang­ing the treated group and con­trol group, we shift the rail­way speed-up car­ried out to an ear­lier data, as 3 to 5 years. After con­trol­ling for in­dus­try ef­fect, time ef­fect and re­gion

ef­fect, Ta­ble 7 shows the re­gres­sion re­sults of placebo test. Ob­vi­ously, there are slight dif­fer­ences in the co­ef­fi­cients of CRH×pe­riod in var­i­ous mod­els of hy­po­thet­i­cal im­ple­men­ta­tion year of rail­way speedup. How­ever, such dif­fer­ences re­main in­signif­i­cant at 10% sta­tis­ti­cal level. These re­sults ex­clude the pos­si­bil­ity of any im­pacts of fac­tors other than rail­way speed-up on the TFP of treat­ment-group firms, and re­flect our core con­clu­sions are ro­bust.

(2) Test based on panel SFA method. We re­cal­cu­late firm’s TFP us­ing panel sto­chas­tic fron­tier anal­y­sis ( SFA) in stead of DEA, and retest the ef­fectS of rail­way speed- up on firm’s TFP by the dif­fer­ence-in-dif­fer­ences method. Ta­ble 8 re­ports the re­gres­sion re­sults. In Ta­ble 8, the co­ef­fi­cient of CRH×pe­riod for all sam­ples is sig­nif­i­cantly pos­i­tive, which shows that rail­way speed-up has pro­moted the TFP of firms along the route and ver­i­fies the core con­clu­sions above. Sam­ple-spe­cific re­gres­sion re­sults show that the CRH×pe­riod is highly sig­nif­i­cant and pos­i­tive for non-SOEs, firms in coastal re­gions and ex­port firms; more­over. This re­sult co­in­cides with our pre­vi­ous con­clu­sions. Based on the above re­sults, we may find that re­mea­sur­ing firm’s TFP by a new method will not cause any sig­nif­i­cant im­pacts on our key con­clu­sions.

(3) Test based on fi­nan­cial data. The above sec­tion es­ti­mates the tech­nol­ogy and ef­fi­ciency of firms us­ing the data en­vel­op­ment anal­y­sis (DEA) method. To fur­ther test the ef­fects of rail­way speed-up on com­mu­nity busi­nesses, we con­duct re­gres­sion us­ing the new prod­uct out­put and fi­nan­cial data, such as general and ad­min­is­tra­tive ex­penses of firms in Ta­ble 9. Col­umn 2 of Ta­ble 9 shows the re­gres­sion re­sults of all sam­ples. Judg­ing by the re­gres­sion co­ef­fi­cient of CRH×pe­riod, rail­way speed-up leads to a sig­nif­i­cant im­prove­ment in the firm’s new prod­uct out­put value. In­creased mar­ket com­pe­ti­tion re­sult­ing from short­ened tem­po­ral and spa­tial rail­way dis­tance may in­duce firms to adopt new tech­nol­ogy for new prod­uct man­u­fac­tur­ing or de­velop new prod­ucts us­ing new de­sign ap­proaches, which leads to an in­crease in new prod­uct value. A sam­ple-spe­cific re­gres­sion re­sult is con­sis­tent with the above key con­clu­sions. Namely, rail­way speed-up has a sig­nif­i­cant ef­fect on the new prod­uct value of pri­vate firms in coastal re­gions; how­ever, it has an in­signif­i­cant im­pact on SOEs in in­land re­gions, but the dif­fer­ence is also in­signif­i­cant be­tween non-ex­port firms and ex­port firms. Mean­while, to­tal sam­ple re­gres­sion re­sults show that rail­way speed-up sig­nif­i­cantly re­duced cor­po­rate man­age­rial cost. How­ever, con­sid­er­a­tion of sam­ple het­ero­gene­ity leads to a dif­fer­ent con­clu­sion, i.e. rail­way speed-up sig­nif­i­cantly re­duces the general and ad­min­is­tra­tive cost of pri­vate firms, firms in coastal re­gions and ex­port firms, but the ef­fect is in­signif­i­cant or nega­tive for state-con­trol en­ter­prises, firms in in­land re­gions and non-ex­port firms. The level of general and ad­min­is­tra­tive cost re­flects the level of a com­pany’s man­age­rial ef­fi­ciency and cor­po­rate ef­fi­ciency. These find­ings sup­port the core con­clu­tions of this paper.

4.4 Fur­ther Dis­cus­sion: In­fer­ences Based on Ge­o­graph­i­cal Spheres

This paper clas­si­fies firms into dif­fer­ent ge­o­graph­i­cal spheres based on their dis­tance to high-tech cities like Beijing, Shang­hai, Guangzhou and Shen­zhen to test the het­ero­ge­neous ef­fects of rail­way speed-up on the firm’s TFP within dif­fer­ent spheres, and the test re­sults are shown in Ta­ble 10. No­tably, rail­way speed-up has the most sig­nif­i­cant ef­fects on the mtc of firms within a dis­tance of around 150 kilo­me­ters to high- tech cities like Beijing and Shang­hai. Given the EMU train op­er­a­tion speed is be­tween 160 to 200 km/h and de­duct­ing the time of train stops at sta­tions, we may get the fol­low­ing con­clu­sion: Rail­way speed- up ex­panded “one- hour eco­nomic cir­cle” from 100 km de­ter­mined by or­di­nary rail­way or high­way to 150 km. Ag­glom­er­a­tion and com­pe­ti­tion ef­fects of “one-hour eco­nomic cir­cle” pro­moted cor­po­rate de­vel­op­ment within this sphere. Sam­ple-spe­cific re­gres­sion re­sults show that rail­way speed-up has sig­nif­i­cant ef­fects on the tech­nol­ogy of pri­vate and ex­port-ori­ented firms . Tech­nol­ogy progress of firms ben­e­fits from the “one-hour eco­nomic cir­cles” of cen­tral cities con­nected through high-speed rail­way or metro. Alarm­ingly, the mtec of firms does not ex­hibit a pos­i­tive ef­fect con­sis­tent with mtc. This im­plies that while rail­way speed-up pro­motes cor­po­rate tech­nol­ogy, it can also bring about some nega­tive ef­fects on firms, es­pe­cially in terms of ef­fi­ciency im­prove­ment..

5. Con­clu­tions

Rail­way speed- up is an im­por­tant event for China’s trans­porta­tion in­fra­struc­ture qual­ity im­prove­ment. Specif­i­cally, the EMU trains launched dur­ing rail­way speed-up and the rapid de­vel­op­ment of high-speed trains are mile­stones in the his­tory of China’s rail­way de­vel­op­ment, and pro­foundly

in­flu­enced its eco­nomic de­vel­op­ment trend over re­cent years. This paper re­gards rail­way speed-up as a quasi-nat­u­ral ex­per­i­ment in China’s trans­porta­tion in­fra­struc­ture qual­ity im­prove­ment. With EMU train sta­tions as group­ing cri­te­rion, we es­ti­mates the firm’s TFP us­ing the data en­vel­op­ment method and then ex­am­ines the TFP ef­fects of rail­way speed-up us­ing the dif­fer­ence-in-dif­fer­ences method. Fi­nally, we fur­ther de­com­pos­es­the TFP into two in­flu­ence fac­tors as tech­nol­ogy change in­dex and ef­fi­ciency change in­dex. Our study re­sults re­veal that rail­way speed-up has sig­nif­i­cantly pro­moted the tech­nol­ogy of firms in rel­e­vant re­gions, but its ef­fi­ciency im­prove­ment ef­fect is in­signif­i­cantly pos­i­tive. Their cou­pled ef­fect has pro­moted cor­po­rate TFP growth and is of more pos­i­tive sig­nif­i­cance for pri­vate firms, firms in coastal re­gions and ex­port firms. De­spite the dif­fer­ent ef­fects of rail­way speed-up on cor­po­rate TFP due to het­ero­gene­ity of re­gions and firms, these dif­fer­ent ef­fects do not cre­ate any sig­nif­i­cant shocks to our core con­clu­sions. Ac­cord­ing to the re­gres­sion re­sult based on the ge­o­graph­i­cal sphere of firms de­ter­mined by their dis­tance to high-tech cities, rail­way speed-up has the most sig­nif­i­cant ef­fect within the “one-hour eco­nomic cir­cle”.

De­spite high costs of rail­way speed-up and high-speed rail­way con­struc­tion and crit­i­cism un­der the fis­cal bud­get con­straint, rail­way in­fra­struc­ture qual­ity im­prove­ment has pos­i­tive and suf­fi­ciently ver­i­fied ef­fects on macroe­co­nomic de­vel­op­ment and cor­po­rate TFP. Rail­way in­fra­struc­ture qual­ity im­prove­ment re­mark­ably shorten the real dis­tance be­tween re­gions, ac­cel­er­ate the cross-re­gional flows of pas­sen­gers, lo­gis­tics and in­for­ma­tion, re­source al­lo­ca­tion, and in­crease mar­ket com­pe­ti­tion. Cur­rently, sig­nif­i­cant dis­par­i­ties in trans­porta­tion in­fra­struc­ture still ex­ist across var­i­ous re­gions in China. Rail­way speed-up and high-speed rail­way con­struc­tion re­main con­cen­trated in China’s eastern and cen­tral re­gions. As a re­sult, not all re­gions in China can ben­e­fit from rail­way speed-up. In fu­ture plan­ning, at­ten­tion should be given to the qual­ity of rail­way in­fra­struc­ture in China’s less de­vel­oped re­gions. Rea­son­able plan­ning and in­fra­struc­ture con­struc­tion guid­ance should be of­fered to pro­mote the pos­i­tive ef­fect of in­fra­struc­ture qual­ity im­prove­ment and re­duce de­vel­op­ment dis­par­i­ties across re­gions. Con­sid­er­ing the pos­i­tive sig­nif­i­cance of the “one-hour eco­nomic cir­cle” for pro­duc­tiv­ity growth, pri­or­ity should be given to the de­vel­op­ment and im­prove­ment of high-speed ded­i­cated pas­sen­ger lines and in­ter­city rail­ways to en­hance the spillover ef­fects of cen­tral cities on neigh­bor­ing cities.

Newspapers in English

Newspapers from China

© PressReader. All rights reserved.