The Role of Infrasrucure Invesmen Locaion in China s Wesern Developmen By Xubei Luo Developmen of he wesern region is vial o he balanced growh of China. Luo sudies he impacs of infrasrucure invesmen ha may mos efficienly alleviae he burden of geographical remoeness of he Wes. Having consruced he adjused disance o approximae he ranspor cos, which akes ino accoun he effecs of real disance and infrasrucure developmen, he auhor defines he peripheral degree o measure he effecive remoeness of a province o an economic cener. Using panel daa for 1979-99 from he Chinese provinces, she shows ha geographic araciveness plays a significan role in a Solow-ype growh deerminaion model. Given he invariabiliy of pure geographic posiion, progress in ransporaion faciliies is essenial o reduce he geographic handicap and o encourage he caching-up of he wesern region. The auhor s simulaion resuls show ha he cenral ransporaion hubs (Hube Henan, and Hunan) meri mos he infrasrucure invesmens, for hey favor he developmen of many provinces, if regional balanced growh is considered as he prime objecive. In paricular, improvemen in he ransporaion faciliies in cenral hubs will have greaer effecs on wesern developmen han ha in he wesern region by iself. Improvemens in he ransporaion faciliies of he cenral hubs subsanially improves he geographic araciveness of he wesern region by reducing he ranspor cos from he Wes o he Coas and by promoing he emergence of new economic ceners in such hubs, which ends o modify he naional economic geographic srucure. World Bank Working Paper 3345, June 2004 The Policy Research Working Paper Series disseminaes he findings of work in progress o encourage he exchange of ideas abou developmen issues. An objecive of he series is o ge he findings ou quickly, even if he presenaions are less han fully polished. The papers carry he names of he auhors and should be cied accordingly. The findings, inerpreaions, and conclusions expressed in his paper are enirely hose of he auhors. They do no necessarily represen he view of he World Bank, is Execuive Direcors, or he counries hey represen. Policy Research Working Papers are available online a hp://econ.worldbank.org. Xubei Luo The World Bank Office of Senior Vice Presidency Dev Econ/Chief Econ (DECVP) MSN MC4-404 1818 H Sree, N.W. Washingon D.C. 20433 Tel: 1 202 458 1157 Email: xluo@worldbank.org
1999. 1 Despie his grea success of he economy as a whole, China has suffered from The Role of Infrasrucure Invesmen Locaion in China s Wesern Developmen By Xubei Luo * 1. Inroducion China, wih a populaion of 1.3 billion and a surface of 9.6 million squared kilomeers, has achieved grea success since he lae 1970s. The counry s economic size increased more han 6.5 imes in 21 years as he Gross Domesic Produc (GDP) increased from $145 billion USD (consan 1995) in 1978 o $964 billion USD (consan 1995) in unbalanced growh among he differen provinces: he coasal region is much more developed han he inland area due o more favorable geographic posiion and economic policies. 2 In general, he farher away a province is from he coas, he weaker is performance is. In 1999, he GDP per capia of he wesern provinces was less han half of ha of he coasal provinces. Such unbalanced regional growh hinders he furher developmen of he Chinese economy. On one hand, he backwardness of he wesern region and resuling inequaliy of incomes across China undermines longer-erm growh prospecs and social sabiliy; on he oher hand, he underdevelopmen limis he poenial of domesic marke enlargemen, hindering he possibiliy of he relocaion process of radiional indusries from easern regions o cenral and wesern ones. In addiion o he opological and climaic disadvanages, he wesern provinces experience limied marke accessibiliy due o heir geographical remoeness from he domesic economic ceners of he coas. We argue ha i is he ranspor cos, which is * The auhor is indebed o Deepak Bhaasali and Paul Brenon for excellen commens. The auhor is also graeful o Henri-François Henner and Nong Zhu for fruiful discussion. 1 Source: World Bank daabase. 2 In general, according o he crierion of geographic posiion, we classify Chinese provinces ino hree caegories: coasal, cenral, and wes. According o he regional classificaion of he "Saisics of China's Fixed Asses and Invesmen 1950-1985," he coasal region includes Liaoning, Hebe Beijing, Tianjin, Shandong, Jiangsu, Shangha Zhejiang, Fujian, Guangdong, Hainan; he cenral region includes Heilongjiang, Jilin, Nei Mongol, Shanx Henan, Anhu Hube Hunan, Jiangxi; he wesern region includes Guangx Shaanx Gansu, Ningxia, Qingha Xinjiang, Chongqing, Sichuan, Yunnan, Guizhou, Xizang. Here, alhough Guangxi has direc access o sea, i is considered as a wesern province given is economic backwardness. The cenral region and he wesern region are considered as he inland regions. See annex 1 for more deails. 1
condiioned by he real disance and by he infrasrucure developmen level of he iinerary, raher han he disance per se ha influences inerregional rade volume. We hus consruc an indicaor "peripheral degree" o measure he geographic araciveness of a province, namely, is effecive remoeness o economic ceners. The significan negaive role ha he remoeness o he economic ceners, ha is, he peripheral degree, plays on regional growh suggess a pessimisic prognosis for he developmen of he wesern provinces. The mos efficien way o faciliae he growh of he inland/wesern provinces is o develop he infrasrucure o lower he ranspor cos and lessen he relaive effecive remoeness of he wesern region. In general, for a given disance, he beer he infrasrucure is, he lower he ranspor coss are and he beer he marke accessibiliy is. Infrasrucure developmen will shoren he disance, so ha geographic posiion will play a less imporan role. Given curren limiaions on infrasrucure invesmens, he key quesion becomes how o efficienly locae infrasrucure invesmen o opimize he effecs on regional developmen. By simulaing he influence of a 10 percen increase in ransporaion nework densiy in each province, we demonsrae ha infrasrucure invesmens in ransporaion hubs in cenral China have greaer effecs on wesern developmen han hose in he wesern region; he former lowers he effecive remoeness of he Wes o economic ceners. On he one hand, he improvemen of he ransporaion faciliies in cenral hubs reduces he ranspor cos beween Wes and Eas; on he oher hand, he emergence of he cenral provinces as regional economic ceners modifies he geo-economic producion srucure in favor of lowering he peripheral degree of he Wes. 3 This paper is organized as follows: secion wo emphasizes he unbalanced regional growh and briefly reviews he heoreical models; secion hree focuses on he consrucion of he adjused disance and peripheral degree; secion four demonsraes he role of geographic posiion on regional growh deerminaion; secion five simulaes he effecs of differen infrasrucure invesmen locaion on regional growh; and secion six presens our policy suggesions on he developmen of ransporaion faciliy in cenral ransporaion hubs. 2. Background Descripion and Theoreical Review In China, coasal provinces have experienced higher growh raes and beer developmen han have he inland provinces. Accordingly, we consider he coasal provinces 3 See Fujia, Krugman, and Venables (1999). 2
o be economic ceners. Favored by economic policies of opening-up and by advanageous geographical posiions, he growh rae of coasal provinces was much higher han ha of he inland provinces for he las wo decades. This higher growh rae furher enlarged he developmen gap (able 1). 4 Table 1. Provincial GDP Per Capia and Annual Growh Rae Provincial GDP per capia (yuan 1978) Average annual growh rae (%) Code Province 1978 1985 1992 1999 79-89 90-99 79-99 Coasal Region 1 Beijing 1249 2109 3358 6222 7.01 8.35 7.65 2 Tianjin 1141 1941 2714 5656 6.14 9.25 7.62 3 Hebei 362 586 994 2239 6.99 10.53 8.68 6 Liaoning 675 1107 1759 3164 7.23 7.49 7.35 9 Shanghai 2484 4032 6161 14065 6.16 10.57 8.26 10 Jiangsu 427 872 1722 4082 9.94 11.64 10.75 11 Zhejiang 330 805 1529 3796 10.69 12.68 11.63 13 Fujian 271 567 1102 2895 9.73 13.00 11.29 15 Shandong 315 614 1099 2614 8.67 11.62 10.08 19 Guangdong 365 725 1637 3600 10.06 11.83 10.90 Cenral Region 4 Shanxi 363 637 905 1611 6.70 7.53 7.10 5 Nei Mongol. 318 630 964 1747 8.16 8.04 8.11 7 Jilin 381 668 1099 2115 8.03 8.29 8.16 8 Heilongjiang 558 835 1218 2102 5.62 7.07 6.31 12 Anhui 242 483 657 1548 8.07 9.68 8.84 14 Jiangxi 273 488 781 1623 7.51 9.55 8.48 16 Henan 231 453 702 1480 8.34 9.42 8.85 17 Hubei 330 662 973 2108 8.14 9.59 8.83 18 Hunan 285 463 702 1363 6.38 8.65 7.46 Wesern Region 20 Guangxi 223 334 534 1062 5.15 9.95 7.44 21 Sichuan 252 475 737 1361 7.73 8.36 8.03 22 Guizhou 174 326 472 765 7.74 6.33 7.07 23 Yunnan 223 395 650 1141 7.88 7.64 7.77 24 Shaanxi 292 500 782 1415 7.81 7.20 7.52 25 Gansu 346 515 844 1487 6.69 7.22 6.94 26 Qinghai 426 627 805 1355 4.81 6.29 5.51 27 Ningxia 366 618 885 1531 7.07 6.54 6.82 28 Xinjiang 317 621 1098 1769 8.70 7.63 8.19 Source: China Saisical Yearbook (various ediions) and auhor's calculaions. The daa of Chongqing are included in Sichuan for beer coherence. We do no include Hainan, which is an island wihou direc road/railway connecion wih he coninen, and Xizang because of heir special characerisics. 4 See Démurger and ohers (2001) for a deailed descripion of Chinese regional developmen hisory favored by he geographically argeed Region Developmen Plan iniiaed under Deng Xiaoping and he advanageous endowmen, coasal provinces developed much beer han he inland ones following reforms. 3
Taking ou he hree ciy provinces of Beijing, Tianjin, and Shangha figure 1 shows he enlarging regional developmen gap: a he end of 1990s, all coasal provinces were more developed han he inland ones. 5 The curves parallel o he axis year show he beer performance of he coasal provinces during he las wo decades wih he seeper slopes; hose parallel o he axis code confirm he backwardness of he inland provinces, in paricular he wesern ones, wih lower levels of GDP per capia in each corresponding year. Coasal Provinces Inland Provinces 45 Real GDP Per Capia (100 yuans 1978) 40 35 30 25 20 15 10 5 0 1011 19 6 13 15 3 7 17 8 28 5 14 4 12 27 25 16 24 18 21 26 23 20 22 Provincial Code 1998 1994 1990 1986 1982 1978 Year Figure 1. Enlarging Developmen Gap beween Coasal Provinces (excep Beijing, Tianjin, and Shanghai) and Inland Provinces, 1978-99 I is generally rue in China as well as hroughou he world ha disance from economic ceners is associaed wih he limied marke access. The influx of goods and echnologies become more expensive and he exporaions o parners gain less prof for he burden of ranspor coss increases wih disance, oher hings being equal (Venables and Limao 1999). The geographical araciveness of a region, measured by is marke access, is hence deermined by is relaive remoeness o he economic cener. The greaer he marke access is, he higher he economy of scale is ha an economy can pursue. The graviy model developed by Bergsrand (1985), Deardorff (1995), and Guillaumon, Brun, and De Melo (1998), o name a few, demonsraes ha he greaer he 5 The hree municipaliies direcly under he Cenral Governmen Beijing, Tianjin, and Shanghai always have a much higher level of real GDP per capia han he oher provinces. The real GDP per capia of Shanghai is a leas five imes he naional average level, and ha of Beijing and Tianjin is wice more. 4
economic mass of he rade parners and he smaller he disance beween hem, he greaer heir bilaeral rade volume is. More ofen, he GDP level is aken as he measure of economic mass, he aracion facor in graviy model. However, as o he resisance facor, real disance beween wo concerning regions is no necessarily he bes proxy (Hummels 1999a, 1999b; Walz 1996; Crafs and Venables 2001). I is no he disance per se ha hinders bilaeral commerce; raher, i is he ranspor cos, which lessens he profi margin beween cos of exporer and price of imporer, ha impedes rade. As developed in Limao and Venables (2000), ranspor cos is deermined no only by he absolue geographical disance beween rade parners bu also by he developmen level of infrasrucures of he iinerary concerned, namely, including ha of he ransi regions. The developmen level of he ransporaion infrasrucure is hus vial for regional developmen. The densiy of ranspor nework is one of he imporan indices of infrasrucure developmen, which may serve as measure of he infrasrucure developmen level (Bro and Gérardin 2001; Cain, Ghio, and von Huffel 2002). The denser he ranspor nework beween wo parners is, he higher he probabiliy is ha a beer roue for ranspor purposes can be idenified. In he absence of ariffs on inernal rade, ranspor cos capures mos of he barriers of inerprovincial rade. For a given real disance, he beer he infrasrucure developmen of iinerary is, he lower he ranspor cos is. Luo (2001) shows ha he disance adjused by infrasrucure developmen level serves as a beer proxy of resisance facor. Growh heory suggess ha regional growh rae is condiioned by is iniial developmen level. If he iniially poorer always ends o grow faser han he iniially richer, i is a case of absolue convergence. However, oher han he iniial developmen level, regions may be heerogeneous in many oher aspecs, such as invesmen rae and demographic growh rae (Solow 1956; Swan 1956; Barro 1991; Barro and Sala-i-Marin 1991, 1992a, 1992b, 1996). I may no be reasonable o consider ha all Chinese regions have he same developmen capaciy in he long run. The heory of condiional convergence, which emphasizes he regional long-erm developmen difference and is deerminaions, corresponds more closely o he realiy: he farher away an economy is from is own longerm developmen capaciy, he faser is growh rae is (Bernard and Durlauf 1995; Ben- David 1997; Venables 1999; Ben-David and Kimhi 2000; Maureh 2001). In oher words, an economy converges o is own seady sae, which is no necessarily common for differen regions (Funke and Srulik 1999). 5
Therefore, he focal poin urns o find ou he facors ha condiion long-erm developmen capaciy. As suggesed by Gallup, Sachs, and Mellinger (1999); Redding and Venables (2000); and Radele and Sachs (1999), high ranspor coss will damage expor performance and economic growh. This paper argues ha he geographical posiion of a region, which influences is marke access, consiues an imporan deerminan of regional economic performance. We will es he role of he effecive remoeness of a region o he economic cener in he Solow-ype growh deerminaion model and examine he effecs of he locaion of infrasrucure invesmen on regional growh o idenify he mos efficien way o favor wesern developmen. 3. Peripheral Degree Geographically adjoining regions end o share similar economic performance. Given he fac ha climae and opology play an imporan role in influencing disease burdens, agriculural produciviy, and locaion in ranspor coss, i is no surprising ha he disribuion of economic aciviies is spaially uneven (Gallup, Sachs, and Mellinger 1999; Henderson, Shaliz and Venables 2000). The homogeneiy of he adjacen regions demonsraes ha geographical posiion may be an imporan facor ha condiions economic growh, as deailed in Baumon and ohers (2000) and Davis and Weinsein (1997). This secion concenraes on consrucing a peripheral degree o measure he relaive remoeness of he province in quesion o an economic cener and examines he role of peripheral degree in regional growh. To measure he marke accessibiliy of a province o a domesic economic cener (here, he coasal provinces), aking ino accoun he relaive imporance of he economic size of differen coasal provinces, we define is peripheral degree, noed as PD, as he weighed sum of he adjused disance beween his province and all he coasal provinces. 6 The influence of he demand from a parner province on he economic performance of he province in quesion is posiively correlaed o he economic mass of he former. We weigh he adjused disance beween he province in quesion (i ) and a coasal province ( j ) by he economic mass of he laer relaive o ha of coasal provinces in oal o approximae he relaive imporance of he effecive remoeness. The reason lies in ha he farher away he province in quesion is from a grea size coasal province, he more serious he disadvanage is ha i suffers. We suppose ha, oher hings being equal, he srucure of demand is similar 6
among differen regions, namely, he percenage of demand saisfied by local producion (and hus ha saisfied by he producion of oher provinces) is he same, as suggesed by Courcier and Laffay (1972). Taking he influence of he economic mass of various coasal provinces and ha of he ranspor cos on marke accessibiliy of he province in quesion ino accoun, we define our peripheral degree as follows: PD = ( DisA j ij, GDP j j, GDP j, ) where PD i, represens he peripheral degree of province i a ime ; DisA ij, represens he real disance beween province i and province j adjused by he developmen level of infrasrucure of he iinerary ha connecs hese wo provinces a ime, namely he adjused disance beween province i and province j a ime suggesed by Luo (2001) 7 ; GDP j, represens he real GDP of province j a ime. Here, j represens he coasal provinces, including Beijing, Tianjin, Hebe Liaoning, Shangha Jiangsu, Zhejiang, Fujian, Shangdong, and Guangdong. 8 Figure 2. Relaion beween Peripheral Degree and Economic Growh Rae of Chinese Provinces, 1978-99 6 In he case of China, all imporan harbors siuae in coasal provinces. The relaive remoeness o coasal provinces is posiively correlaed o he relaive remoeness o inernaional markes. 7 See annex 2 for he consrucion of he adjused disance. 8 We exclude Hainan from our sample for is unique island characerisics. 7
Figure 2 shows a srong correlaion beween peripheral degree and economic performance (in logarihmic forms): he higher he former is, he weaker he laer is. The imporan role of geographic posiion in regional developmen is also confirmed by he correlaion sudy: he rank correlaion beween peripheral degree and real per capia GDP, boh in logarihmic forms, is significanly negaive wih an average Spearman's Rho value of 0.5549, and he probabiliy of independence beween hese wo variables is zero. 9 By examining he Spearman's Rho value of each year, we find ha his rank correlaion becomes more and more significan as ime passes (figure 3). 0,2 0,1 Pr and Spearman s Rho 0,0-0,1-0,2-0,3-0,4-0,5-0,6-0,7 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 Pr> Spearman's Rho Pr> sands for he probabiliy of independence of ldp and ly. Figure 3. Spearman's Rank Correlaion beween Peripheral Degree and Real Per Capia GDP of Chinese Provinces, 1978-99 We argue ha geographical posiion plays an increasingly imporan role on regional developmen, as Chinese economy becomes more marke-oriened. A he beginning of he economic reform in he lae 1970s, he performance of he Chinese economy was disored by he inefficien allocaion of resources, and he regional developmen level did no well reflec 9 To lessen he sensiiviy o exreme values, we prefer Spearman's rank correlaion esimaion o sandard Pearson correlaion esimaion. Spearman's rank correlaion is calculaed as Pearson's correlaion compued on he ranks and average ranks (Spearman (1904), Conover (1980)). The significance is calculaed using he 2 approximaion p = prob( n 2, ρ n 2 / 1 ρ ). 8
he developmen capaciy. 10 As he economic reform deepens, regional comparaive advanages become an imporan facor ha deermines he producion srucure. Favored by beer access o foreign markes and benefied by he opening-up policies, coasal provinces grow much faser han inland ones, exacerbaing he problem of unbalanced growh (Woo 1998). Given he fac ha he geographical economic posiion of a province will no be alered significanly by fundamenal exernal policy change, i is reasonable o predic ha he core peripheral developmen paern will coninue. Appropriae policies may be needed o rebalance regional developmen. 4. Growh Deerminaion Using he macroeconomic daa from Chinese Saisic Yearbooks, Comprehensive Saisic Daa and Maerials on 50 Years of New China, and China Regional Economy: A Profile of 17 Years of Reform and Opening Up, and he disance daa from he Map of Communicaion Faciliies of China, his secion sudies he deerminaion of annual growh of Chinese provinces during 1978-99, emphasizing he role of geographical posiion. The neoclassical growh model suggesed by Solow-Swan (1956) emphasizes he effecs of physical invesmen rae and demographic growh rae. The higher he raio is of physical invesmen o GDP, he higher he produciviy of effecive labor is in long-run equilibrium, oher hings being equal. Trying o explain he enormous conribuion of fixed capial in growh, he augmened Solow-Swan model argues he imporance of human capial. The imporance of he increasing reurn o scale renders he homogeneiy of producion funcion in various regions quesionable, as suggesed by he heory of new economic geography (Krugman 1991). Specializaion and agglomeraion amplify regional differences. 11 Oher han he demographic growh rae, he Solow-Swan model argues ha he echnological progress rae and depreciaion rae of fixed invesmens influence regional growh. However, no all parameers ha condiion long-erm economic performance are available or measurable. The omission of regional specific characerisics may lead o bias in esimaion resuls. Hence, panel daa esimaion is preferable o cross-secion esimaion, for he former may capure he nonmeasurable specific characerisics in he fixed-effec model. We also inroduce year dummies o conrol for he shor-erm effecs. 10 Pursuing he sake of "poliical securiy" and being aware of he "criical hrea from he foreign capialiss," he Chinese governmen allocaed a grea deal of is indusrial invesmen in he inland provinces during is "planned-economy" period. Subordinaed o he "poliical goal," he comparaive advanages of various provinces were no well considered before economic reform. 11 See Ricci 1999, Krugman (1995), Puga (1999), Quah (2001), and Venables (2000). 9
We sar by esing he very simple hypohesis of absolue convergence (divergence) and esimae he following panel model: ln( 1998 y ) ln( y 1) = α+ ρln( y 1) + χdummy + ε = 1979 (1) The firs equaion in able 2 shows ha he iniial developmen level ln( y 1) plays a significan role on regional growh. However, bearing in mind ha Chinese provinces are heerogeneous, we do no hink i reasonable o admi he hypohesis of heir long-erm seady sae homogeneiy he premise of he exisence of absolue convergence (divergence). As we have shown in he preceding secion ha geographical posiion, which represens regional marke accessibiliy, may play an imporan role in influencing economic performance, we inroduce he variable of peripheral degree, noed as PD, ino our esimaion: ln( 1998 y ) ln( y 1) = α+ ρln( y 1) + θln( PDi, ) + χdummy + ε = 1979 (2) In he second equaion, he variable ln( y 1) coninues o be significanly negaive, which signifies condiional convergence. As prediced, he variable PD plays a negaive role on regional growh. The omission of his perinen variable biases esimaion resuls. According o he Solow-Swan model, we suppose ha producion funcion is of he ype of Cobb-Douglas: ln( y ) ln( y 1) = ϑ + υ ln( y 1) + γ ln( s ) + η ln( n + g + δ ) + ε where s sands for he raio of physical invesmen o GDP; n he demographic growh rae; g he echnology progress, and δ he depreciaion rae of physical invesmen. Many preceden sudies on regional growh, such as ha of Mankiw, Romer, and Weil (1992), suppose ha he sum of hese wo parameers is homogeneously equal o 0.05. However, in he case of China, i seems ha his hypohesis is quesionable given he fac ha Chinese provinces are so differen from one anoher. Due o he daa unavailabiliy of g and δ, we prefer o leave heir influences implicily ino fixed effecs: ln( y 1998 ) ln( y 1) = α + ρln( y 1) + θln( PDi, ) + γ ln( s ) + ζ ln( n ) + χdummy + ε (3) = 1979 10
The significance of he demographic growh rae in he hird equaion is consisen o he predicion of growh heory. The nonsignificance of he invesmen raio may be he resuls of he fac ha economic profi opimizaion, represened by marginal reurn equalizaion, may be sacrificed for regional developmen goals; he locaion of he physical invesmen does no respond well o marke demand. Given he significan difference beween coasal and inland regions, according o he Chow es, we esimae he wo subsamples separaely in equaions 4 and 5. As figure 2 shows, he relaionship beween peripheral degree and growh rae is no linear. To furher invesigae he role of peripheral degree, we add wo supplemenary variables, [ ln( ) ] 2 ln( PD ) ln( y 1) ino our model: PD and α + β ln( y 1) + θ ln( PD ) + γ ln( s ) + η ln( n ) ln( y i, ) ln( y 1) = 2 + ζ ln[( PD )] + β [ln( PD ) ln( y )] + 1998 = 1979 χ dummy + ε 1 1 (4) The sixh equaion shows ha for he coasal provinces he variable [ ln( ) ] 2 PD plays a significanly posiive role on regional growh, which implies ha he higher he peripheral degree is, he weaker he growh rae is, while such negaive effecs decrease as is value increases. 12 The ineracive erm ln( PD ) ln( y 1) is also significanly posiive, which signifies ha for a given level of peripheral degree, he higher he iniial developmen level is, he higher he regional growh rae is. According o he classical growh model, as suggesed by he firs equaion, he convergence effecs are capured by he coefficien of he variable ln( y 1) : [ ln( y ) ln( y )] ln( y 1 1 ) = β However, once we have conrolled for he role of peripheral degree, as suggesed by he sixh equaion, he effecs of convergence become: [ ln( y ) ln( y )] ln( y 1 ) 1 = β β ln( PD ). + 1 12 The economeric resuls of he sixh equaion vividly corroboraes our figure 2, when he peripheral degree is modes, he negaive effec is grea; when i is very high, furher increase of geographic remoeness will lead o less economic performance loss. 11
In oher words, he size of convergence effecs is condiioned by he peripheral degree. The posiive sign of he coefficien β 1 and he negaive sign of he coefficien β demonsrae ha convergence effecs decrease as we conrol for he geographical economic posiion and he geographic remoeness limis convergence of he coasal province, oher hings being equal. For he subsample of he inland provinces, hese wo addiional variables are no ha significan, which implies ha over a cerain limi he negaive role of he geographic remoeness ends o be linear. We hence consider equaions 5 and 6 for he growh deerminaion specificaion for inland and coasal provinces respecively. The convexiy of he peripheral degree on regional growh signifies he rap of remoeness ha wesern provinces suffer. As he famous "liquidiy rap," when he ineres rae is lower han a cerain level, is furher reducion may no be able o induce privae invesmen. When peripheral degree is high, grea improvemen will be needed o encourage local performance. Table 2. Provincial Growh Deerminaion: Role of Peripheral Degree, 1978-99 Dependen variable : ln( y ) ln( y 1) Naional Naional Naional Coas Inland Coas Inland (1) (2) (3) (4) (5) (6) (7) ln( y 1) -0.044*** -0.082*** -0.066*** -0.061*** -0.137*** -1.397*** -0.661** (-3.47) (-5.35) (-4.47) (-2.65) (-5.20) (-3.77) (-2.04) ln( PD ) -0.197*** -0.168*** -0.085-0.254*** -5.275*** -2.187** (-4.26) (-3.56) (-1.44) (-3.11) (-3.49) (-2.13) ln( s ) -0.003 0.005-0.013 0.024-0.013 (-0.30) (0.35) (-1.02) (1.59) (-1.04) ln( n ) -1.066*** -0.747** -0.798* -0.839*** -0.703* (-4.35) (-2.58) (-1.95) (-2.97) (-1.71) ln( PD ) ln( PDi, ) 0.273*** 0.106* (3.40) (1.95) ln( y 1) ln( PDi, ) 0.144*** 0.054 (3.61) (1.58) Consan 0.201*** 2.026*** 1.731*** 1.016* 2.678*** 25.827*** 11.552** (5.24) (4.71) (4.01) (1.88) (3.58) (3.60) (2.37) Adjused R square 0.5025 0.5187 0.5585 0.7192 0.5325 0.7391 0.5412 Observaion number 588 588 568 208 360 208 360 Noe: -sudens are in brackes. * significan in 10 percen ; ** significan in 5 percen ; *** significan in 1 percen. ln( y 1) sands for iniial real GDP per capia in logarihmic form; ln( PDi,) peripheral degree in logarihmic form; ln( s ) physical invesmen raio in logarihmic form; ln( n ) populaion growh rae in logarihmic form. For simpliciy, he ime dummies are no represened in he able. 12
5. Simulaions An improvemen of ransporaion faciliies reduces he peripheral degree of he concerned provinces and favors heir economic performance. Hence, he infrasrucure invesmen policies are imporan o regional balanced growh. The peripheral degree, by consrucion, is deermined by he disance beween he province in quesion and he naional economic cener and by he infrasrucure developmen level of he ransi economies. A similar improvemen of ransporaion faciliies in differen provinces does no lead o he same modificaion of peripheral degree in differen provinces. As shown in he following box, an increase of ransporaion nework densiy in one province does no only lead o he change of is own peripheral degree; i also leads o ha of he oher concerned provinces. Transporaion Nework Densiy and Peripheral Degree D v, = a1 DRv, + a2dfv, The change of he ransporaion nework densiy ( D v, ) in province v leads o he modificaion of he peripheral degree of he concerned provinces i ( PD ). n D i v... v, v= Dij, = 1 n provinces n DisA ij, = Dis D ij, ij, Dis ij,: i j j j Dis ij, = a1 DisRij, + a2disfij, j = GDPj, PD ( DisAij, ) GDPj, j j As o a province i, all modificaion of he ransporaion nework locaes in a province v, which siuaes in is iinerary o he coasal provinces, will lead o a modificaion of is peripheral degree. As o a province v, an improvemen of is ransporaion infrasrucure favors all provinces i ha cross i o join he coasal provinces. The heavier he raffic is ha ransis hrough province v, in which case province v is considered as a ransporaion hub, he more provinces will benefi from is ransporaion faciliy improvemen, and he greaer are is effecs on he reducion of he peripheral degree of he concerned provinces. 13
In his secion, we simulae he effecs of a hypoheical increase of 10 percen in he ransporaion nework densiy in each province, as follows, on he growh of he whole China and of he differen regions, o evaluae he effecs of he locaion of he infrasrucure invesmen on regional balanced growh. Firs, using he value of he peripheral degree ha we have consruced wih he observed value of he iniial developmen level, he physical invesmen rae, and he demographic growh rae of all provinces, and applying he values of he coefficiens ( α, ρ, θ, γ, ζ, χ ) ha we have esimaed in able 2, we calculae he esimaed value of he dependan variable " ln( y ) ln( y 1) ", noed as gaes, using he hird esimaion in able 2: ln( y i, ) ln( y 1) = α + ρ ln( y + 1998 = 1979 1) + θ ln( PD ) + γ ln( s ) + ζ ln( n ) χ dummy + ε Second, we muliply he ransporaion nework densiy of he province v ( v =1,2,3, 28) by 110 percen, so we obain a new value of PD i,, noed as PD (, due o he modificaion of D v,. Third, we pu he new value of PD, ( i ino he equaion in replace of i PD, o calculae he new value of he esimaed variable " ln( y ) ln( y 1) ", noed as gaes (. Fourh, we define gadif ( = gaes( gaes, which sands for he increase of he growh rae of province i in year due o he hypoheical increase of he ransporaion nework densiy in province v. Fifh, o ake ino accoun he difference of he economic size of each province, we calculae he weighed average growh effecs due o he modificaion of gaeff ) ( v : gaeff ( n = [ gadif( n i i PIB PIB ] D v,, noed as The underlying idea lies in he noion ha he larger is he economic size of he province in quesion in comparison wih ha of he whole group, he greaer are he effecs of he improvemen of is economic performance on he growh of he group. 14
Finally, o evaluae he effecs of he change of D v, in each province v on he growh of differen groups of provinces and hence on he regional balanced growh, we calculae he following raios: 1) i all Chinese provinces, gaeff ( sands for he growh effecs on he whole Chinese economy due o a 10 percen increase in he densiy of he ransporaion nework in province v in year, noed as gaeff 0 ( 2) i all coasal provinces, gaeff ( sands for he growh effecs on he coasal region due o a 10% increase in he densiy of he ransporaion nework in province v in year, noed as gaeff 1 ( 3) i all inland provinces, gaeff ) ( v sands for he growh effecs on he inland region due o a 10 percen increase in he densiy of he ransporaion nework in province v in year, noed as gaeff 2 ( 4) i all wesern provinces, gaeff ) ( v sands for he growh effecs on he wesern region due o a 10 percen increase in he densiy of he ransporaion nework in province v in year, noed as gaeff 3 ( For simpliciy, we calculae he following variables o sudy he average induced effecs on he whole period 1978-1999 (able 3): 13 gaeff 0( m= gaeff 1( m= gaeff 2( m= gaeff 3( m= 1999 1979 1999 1979 1999 1979 1999 1979 gaeff0( 21 gaeff1( 21 gaeff2( 21 gaeff3( 21 As he firs column "naional" shows, o generae he larges impacs on he growh of he whole China, we should locae he infrasrucure invesmens in he coasal provinces, in paricular, in Shandong, Jiangsu, and Hebei. Given he economic size of he coasal provinces and heir role as economic ceners, i is no surprising ha he improvemen of he 13 The induced effecs on he growh of differen groups of provinces in each year are available upon reques. 15
ransporaion nework in hese provinces resuls in subsanial effecs on he weighed average growh of China as a whole. However, such favorable growh effecs concenrae mainly in coasal regions, as shown he second column "coasal." 14 The rapid growh in coasal provinces will enlarge he regional developmen gap, even hough his growh paern migh be an efficien way o encourage naional growh. Table 3. Regional Growh Induced by 10 percen Hypoheical Transpor Nework Densiy Increase in Differen Provinces (province in quesion included) Naional Coasal Inland Wes Province gaeff 0 ( m Province gaeff 1( m Province gaeff 2( m Province gaeff 3( m Qinghai 0.03 Shanxi Qinghai 0.07 Nei Mongol Xinjiang 0.03 Nei Mongol Xinjiang 0.07 Jilin Nei Mongol 0.13 Jilin Nei Mongol 0.27 Heilongjiang Ningxia 0.26 Heilongjiang Ningxia 0.57 Xinjiang 0.18 Gansu 0.35 Guangxi Gansu 0.77 Qinghai 0.19 Yunnan 0.39 Sichuan Yunnan 0.87 Ningxia 1.62 Heilongjiang 0.55 Guizhou Heilongjiang 1.24 Tianjin 1.70 Guangxi 0.61 Yunnan Guangxi 1.36 Gansu 2.16 Guizhou 0.69 Shaanxi Guizhou 1.55 Yunnan 2.47 Jilin 1.26 Gansu Tianjin 1.77 Jiangxi 2.61 Shanxi 1.59 Qinghai Beijing 2.71 Beijing 2.72 Shaanxi 1.67 Ningxia... Jilin 2.83 Guangxi 3.88 Sichuan 1.85 Xinjiang Shanxi 3.55 Shanxi 4.05 Tianjin 2.72 Jiangxi 1.04 Shaanxi 3.73 Guizhou 4.41 Jiangxi 3.03 Anhui 1.16 Fujian 3.78 Fujian 4.83 Beijing 4.70 Tianjin 3.49 Sichuan 4.04 Liaoning 5.03 Anhui 5.80 Hubei 3.80 Jiangxi 5.50 Shanghai 6.54 Hunan 7.80 Henan 3.99 Shanghai 6.68 Shandong 7.09 Fujian 8.48 Hunan 4.07 Liaoning 9.84 Zhejiang 7.61 Hubei 9.68 Beijing 6.32 Guangdong 10.23 Jiangsu 7.86 Liaoning 9.74 Shanghai 9.68 Zhejiang 10.39 Hebei 9.40 Henan 10.21 Liaoning 12.28 Jiangsu 10.80 Shaanxi 10.58 Shanghai 10.54 Fujian 13.67 Shandong 10.94 Guangdong 10.79 Guangdong 12.80 Guangdong 14.86 Anhui 11.44 Sichuan 10.86 Zhejiang 14.33 Zhejiang 17.52 Hunan 12.44 Anhui 10.98 Hebei 17.87 Hebei 20.24 Hebei 15.00 Hunan 14.34 Jiangsu 19.05 Jiangsu 25.78 Hubei 16.90 Henan 16.13 Shandong 20.31 Shandong 27.90 Henan 17.67 Hubei 19.89 Noe: Figures are presened in 1/10000. " " sands for he figures inferior o 0.0000005. For example, a 10 percen increase in he ranspor nework densiy in Shandong province leads o a 20.31 10-4 increase of he growh rae of he whole China. Coasal provinces are marked in ialics; ranspor hubs are marked in bold. As he hird column "inland" shows, o simulae he economic performance of he inland provinces, which are acually lagging behind, we should locae he invesmens in he cenral provinces ransporaion hubs of Henan, Hube and Hunan. This locaion choice also maximizes he favorable effecs on he growh of he wesern provinces, as shown in he las column. The improvemen of he ransporaion faciliies in hese cenral provinces leads o he greaes impacs on he rapid growh of he wesern provinces. As he hird column "inland" shows, even if he growh of he coasal provinces is no inegraed in he variable gaeff 2 ( m, he improvemen of he ransporaion nework in cerain coasal provinces, in 14 Here we have even negleced he grea muliplier effec of infrasrucure invesmen on local growh. 16
paricular Hebe generaes greaer induced effecs on he growh of he inland provinces han ha in mos of he oher inland provinces. If we focus on he induced effecs on growh in wesern provinces, as shown in he las column "Wes," he effecs of he invesmens in he cenral provinces of Hube Henan, and Hunan are he larges. In addiion, he induced effecs of he invesmens in some coasal or cenral provinces, such as Guangdong, Anhu and Hebe are greaer han hose in mos of he remoe wesern provinces. To sudy he effecs of an improvemen of ransporaion nework densiy in one province on he growh of he oher provinces and verify he robusness of our simulaion resuls, we calculae he same raios excluding he province in quesion. We calculae n gaeff ( = [ gadif( PIB n ] for all i v o purge he effecs of he invesmen on local i PIBi, growh. 15 i Table 3 bis. Regional Growh Induced by 10 Percen Hypoheical Transpor Nework Densiy Increase in Differen Provinces (province in quesion excluded) Naional Coasal Inland Wes Province gaeff 0 ( m Province gaeff 1( m Province gaeff 2( m Province gaeff 3( m Nei Mongol Shanxi Nei Mongol Nei Mongol Heilongjiang Nei Mongol Heilongjiang Jilin Yunnan Jilin Yunnan Heilongjiang Xinjiang Heilongjiang Xinjiang Yunnan Qinghai 0.02 Guangxi Qinghai 0.04 Xinjiang... Sichuan 0.04 Sichuan... Sichuan 0.09 Qinghai 0.11 Guangxi 0.16 Guizhou... Gansu 0.35 Sichuan 0.34 Gansu 0.16 Yunnan Guangxi 0.37 Gansu 1.03 Ningxia 0.19 Shaanxi Ningxia 0.43 Guangxi 1.17 Guizhou 0.38 Gansu Guizhou 0.86 Ningxia 1.22 Shanxi 0.79 Qinghai Tianjin 1.77 Tianjin 1.70 Jilin 0.80 Ningxia Shanxi 1.82 Jiangxi 2.61 Shaanxi 0.96 Xinjiang Jilin 1.86 Guizhou 2.62 Tianjin 1.73 Jiangxi 1.04 Shaanxi 2.19 Beijing 2.72 Jiangxi 2.14 Anhui 1.16 Beijing 2.71 Shanxi 4.05 Beijing 2.90 Tianjin 1.70 Jiangxi 3.58 Fujian 4.83 Anhui 4.49 Beijing 3.06 Fujian 3.78 Liaoning 5.03 Hunan 6.34 Hubei 3.80 Shanghai 6.68 Shanghai 6.54 Shanghai 6.50 Henan 3.99 Anhui 8.85 Shaanxi 6.88 Fujian 7.39 Hunan 4.07 Hunan 9.41 Shandong 7.09 Hubei 7.75 Shanghai 6.34 Liaoning 9.84 Zhejiang 7.61 Henan 8.01 Liaoning 6.52 Guangdong 10.23 Jiangsu 7.86 Liaoning 8.11 Fujian 10.46 Zhejiang 10.39 Hebei 9.40 Guangdong 10.39 Guangdong 10.52 Jiangsu 10.80 Guangdong 10.79 Zhejiang 12.38 Zhejiang 14.16 Shandong 10.94 Anhui 10.98 Jiangsu 16.36 Hebei 17.85 Hubei 13.13 Hunan 14.34 Hebei 16.48 Jiangsu 21.81 Henan 13.41 Henan 16.13 Shandong 17.48 Shandong 23.63 Hebei 15.00 Hubei 19.89 Noe : Figures are presened in 1/10000. " " sands for he figures inferior o 0.0000005. For example, a 10 percen increase in he ranspor nework densiy of Shandong province leads o a 17.48 10-4 increase of he growh rae of China as a whole (excep Shandong). Coasal provinces are marked in ialics; ranspor hubs are marked in bold. 15 If he province v is no a member of he group in quesion, he value of he raio in able 3 bis is he same as he one in able 3. 17
As shown in able 3 bis, he infrasrucure invesmens in he hree cenral provinces of Henan, Hube and Hunan generae he larges induced effecs on he growh of he wesern provinces, even when he effecs on heir local growh are excluded. Tha he infrasrucure invesmens in he hree cenral provinces of Henan, Hube and Hunan generae he larges induced effecs on he growh of he wesern provinces, even when he effecs on heir local growh are excluded, confirms he vial imporance of he improvemen of ransporaion faciliies in hese provinces, which are densely crossed by raffics. To es he robusness of our simulaion resuls, we use equaions 5 and 6 in able 2 for inland provinces and coasal provinces respecively, which consider he difference beween hese wo regions. 16 Our resuls show ha i is always he improvemen in cenral ransporaion hubs ha generaes he highes effecs on regional balanced growh. 17 6. Regional Policy Suggesions We sugges developing he infrasrucure of he provinces ha are on he way of ransporaion from many wesern provinces o markes, namely, he ransporaion hubs of China. In fac, he marke access of one province, and hence is long-erm developmen, depends no only on is own infrasrucure developmen bu also on he infrasrucure developmen of he ransi provinces. To reduce he effecive remoeness and o enlarge he marke access of he wesern region, we should develop he infrasrucure ha connecs he provinces o markes. Since he governmen finances mos of he infrasrucure invesmens, i is possible o carry hem ou in seleced provinces o favor he Wes. Wih he nonnegligible muliplier effec of infrasrucure invesmen on local economic developmen, he addiional invesmen in he coasal provinces may aggravae he unbalanced developmen of China. Furher infrasrucure developmens wihin coasal regions may lead o he enforcemen of agglomeraion, which will accelerae regional growh of coasal regions relaive o he wesern ones. Such policies violae he equiy crierion. Oher hings being equal, we sugges giving prioriy o he developmen of he wesern region o ensure basic infrasrucure developmen. If he infrasrucure is oo poor for local enerprises o have he minimum demand because of marke segmenaion, he developmen of wesern region may be unrealizable due o he lack of microeconomic foundaions. The 16 See annex 3 for deails. 17 We have also simulaed a 10 percen increase in he naional average ransporaion nework densiy o conrol for he effecs of differen hypoheical densiy increase in absolue value due o he differen iniial densiy of ransporaion nework in each province. The ranking of he provinces regarding o he size of he induced growh effec says he same. Resuls are available upon reques. 18
appropriae invesmen in he Wes is necessary o guaranee basic inraregional ransporaion faciliies. However, while he developmen of he infrasrucure inside he region faciliaes inraregional ransporaion and enlarges he marke of local enerprises, i does so only wihin he province iself. The poliical inclinaion o wesern regional developmen does no necessarily lead o he conclusion ha all infrasrucure invesmens should be locaed in he wesern region. To so locae infrasrucure invesmens would enail he risk ha, wihou a sufficien infrasrucure developmen of he cenral region, rade volume beween he Eas and he Wes would be limied. Because wesern provinces are far more underdeveloped, heir local demands are modes and heir echnology and managemen are less advanced. The separaion of Eas and Wes renders he invesmen in he Wes much less efficien. To ake he wo crieria, equiy and efficiency, ino accoun a he same ime, i is more advisable o locae infrasrucure invesmens in he hubs of Henan, Hube and Hunan. As ransporaion hubs, alhough hey may be less developed compared wih he coasal provinces, hey are already relaively developed, a leas in erms of ransporaion condiions. For he specialy of infrasrucure invesmen iself, he addiional sum o a reasonably welldeveloped locaion generaes greaer posiive effecs han hose o less-developed places. This is no o conclude ha we should neglec he developmen of remoe border provinces because of heir curren relaive less developed siuaions. Developing he hubs favors he ouward-looking producion srucure of he wesern region and faciliaes he ransfer of echnology and managemen skills from Eas o Wes. However, one poenial problem ha we mus consider is ha when we srenghen infrasrucures, say, he ransporaion nework, from he Wes o he Eas, we improve he accessibiliy of he easern provinces o wesern markes. While his improved accessibiliy acceleraes he developmen of he Wes, i has inheren risks insofar as he coasal and cenral regions can supply beer goods a lower prices. For he sake of long-erm growh of he Wes and hus of China as a whole some appropriae regional economic policies, such as regulaions ha reasonably proec he privileges of wesern local enerprises as infan indusries, mus be issued o limi his negaive effec. In addiion, given he fac ha he infrasrucure developmen level of some wesern provinces is oo poor o provide basic faciliies of ransporaion and ha infrasrucure invesmen iself generaes a large muliplier effec on local developmen, i is reasonable o locae some infrasrucure invesmen here o help o break he vicious cycle. In paricular, invesmens in he provinces Sichuan and Shaanxi are recommended, given heir imporan role in regional developmen. 19
In shor, we should ake boh he efficiency crierion and he equiy crierion ino consideraion o choose locaions of infrasrucure invesmens. Because cenral hubs favor he balanced developmen of China, hey are he bes locaions. 7. Conclusions Infrasrucure developmen is imporan for regional performance. Geographic remoeness o economic ceners significanly hinders regional growh. Improving ransporaion faciliies, which reduces ransporaion coss, increases he volume of rade and echnology ransfer. Targeed infrasrucure invesmen in cenral ransporaion hubs favors he caching up of he wesern region mos efficienly by opimizing marke access. References Barro, R.J. 1991. "Economic Growh in a Cross Secion of Counries." Quarerly Journal of Economics 106(2):407-43. Barro, R.J., and X. Sala-i-Marin. 1991. "Convergence across Saes and Regions." Brooking Papers on Economic Aciviy, No.1:107-82. Barro, R.J., and X. Sala-i-Marin. 1992a. "Convergence." Journal of Poliical Economy 100(2): 223-51. Barro R.J., and X. Sala-i-Marin. 1992b. "Regional Growh and Migraion: a Japan-Unied Saes Comparison." Journal of he Japanese and Inernaional Economies 6:312-46. Barro R.J., and X. Sala-i-Marin. 1996. La croissance économique. Collecion sciences économiques, McGraw-Hill /Ediscience, Paris. Baumon C., P. P. Combes, P. H. Derycke, and H. Jaye. éds. 2000. Economie géographique: les héories à l'épreuve des fais. Paris: Economica. Ben-David, D. 1997. "Convergence Clubs and Subsisence Economies." Naional Bureau of Economic Research Working Paper 6267. Cambridge, Mass. Ben-David D., and A. Kimhi. 2000. "Trade and he Rae of Income Convergence." Naional Bureau of Economic Research. Naional Bureau of Economic Research Working Paper 7462. Cambridge, Mass. Bergsrand, J.H. 1985. "The Graviy Equaion in Inernaional Trade: Some Microeconomic Foundaions and Empirical Evidence." Review of Economics and Saisics (U.S.), 67:474-81. Bernard A.B., and S. N. Durlauf. 1991. "Convergence in Inernaional Oupu." Naional Bureau of Economic Research. Working Paper 3717. Cambridge, Mass. Bro J., and H. Gérardin. 2001. Infrasrucure e développemen. Paris: L'Harmaan. Cain M., S. Ghio, and C. van Huffel. 2002. "Infrasrucures de ranspor e inégraion européenne: efficacié économique versus équié régionale." Revue d'economie Poliique, 112(3): 409-36. Conover, W.J. 1980. Pracical Nonparameric Saisics, 2 ème ed. New York: John Wiley & Sons. Courcier, M., and G. Lafay. 1972. "Simulaion économique mulinaionale." Saisiques & Eudes Financières, 1972/8, 4 ème rimesre. Crafs N., and A. J. Venables. 2001. "Globalizaion in Hisory: a Geographical Perspecive." 20
Accessible a hp://www.les.ac.uk/saff/ajv/. Deardorff, A. 1995. "Deerminans of Bilaeral Trade: Does Graviy Work in a Neoclassical World?" Naional Bureau of Economic Research Working Paper 5377. Cambridge, Mass. Davis, D.R., and D. E. Weinsein. 1997. "Economic Geography and Regional Producion Srucure: an Empirical Invesigaion." Naional Bureau of Economic Research Working Paper 6093. Cambridge, Mass. Démurger S., J. D. Sachs, W. T. Woo, and S. Bao. 2002. "Geography, Economic Policy and Regional Developmen in China." Naional Bureau of Economic Research Working Paper 8897. Cambridge, Mass. Fujiaa, M., P. Krugman, and A. J. Venables. (1999. The Spaial Economy: Ciies, Regions and Inernaional Trade." Cambridge, Mass.: MIT Press. Funke, M., and H. Srulik. 1999. "Regional Growh in Wes Germany: Convergence or Divergence." Economic Modelling 16: 489-502. Gallup, J.L., J. D. Sachs, and A. D. Mellinger. 1999. "Geography and Economic Developmen." Inernaional Regional Science Review 22(2):179-232. Guillaumon, P., J. F. Brun, and J. De Melo. 1998. "La disance abolie? Crières e mesure de la mondialisaion du commerce exérieur." Eudes e documens du CERDI E98.30. Henderson, J.V., S. Zhaliz and A. J. Venables. 2000. "Geography and Developmen." World Bank Working Paper 2456. Washingon, D.C. Hummels, D. 1999a. "Toward a Geography of Trade Coss." Accessible a hp://gsbwww.uchicago. edu/ fac/david.hummels/research/vs/. Hummels, D. 1999b. "Have Inernaional Transporaion Coss Declined?" Accessible a hp://gsbwww.uchicago.edu/fac/david.hummels. Krugman, P. 1991. Geography and Trade. Cambridge, Mass: Leuwen Universiy Press and MIT Press. Krugman, P. 1995. "A Dynamic Spaial Model." Naional Bureau of Economic Research Working Paper 4219. Cambridge, Mass. Limao, N., and A. J. Venables. 2001. "Infrasrucure, Geographical Disadvanage and Transpor Coss." World Bank Economic Review (Inernaional) 15(3):451-79. Luo, X. 2001. "La mesure de la disance dans le modèle de gravié une applicaion au commerce bilaéral enre les provinces chinoises avec le Japon." Revue Région e Développemen No.13:163-80 Mankiw, N.G., D. Romer, and D. N. Weil. 1992. "A Conribuion o he Empirics of Economic Growh." Quarerly Journal of Economics 107(2):407-37. Maureh, P.B. 2001. "Convergence, Geography and Technology." Srucural Change and Economic Dynamics 12:247-76. Puga, D. 1999. "The Rise and Fall of Regional Inequaliies." European Economic Review 43:303-34. Radele S., and J. D. Sachs. 1999. "Shipping Coss, Manufacured Expors and Economic Growh." HID Cambridge, Mass. Redding S.J., and A. J. Venables. 2004. "Economic Geography and Inernaional Inequaliy." Journal of Inernaional Economics (Neherlands) 62(1):53-82. Ricc L.A. 1999. "Economic Geography and Comparaive Advanage: Agglomeraion versus Specializaion." European Economic Review 43:357-77. Quah, D. 2001. "Some Simple Arihmeic on How Income Inequaliy and Economic Growh Maer." Accessible a hp://econ.lse.ac.uk/saff/dquah/p/0106iieg.pdf. Solow, R.M. 1956. "A Conribuion o he Theory of Economic Growh." Quarerly Journal of Economics 70(1):65-94. Spearman, C. 1904. "The Proof and Measuremen of Associaion beween Two Things." American Journal of Psychology 15:72-101. Sae Saisical Bureau of he People's Republic of China. China Saisical Yearbook. Beijing: China Saisics Press. Swan, T.W. 1956. "Economic Growh and Capial Accumulaion. Economic Record, 32: 334-61. Venables, A.J. 1999. "Regional Inegraion Agreemens: a Force for Convergence or Divergence." World Bank Working Paper 2260. Washingon, D.C. 21
Venables, A.J. 2000. "Winners and Losers from Regional Inegraion Agreemens." Economic Journal, 113:747-61. Venables, A.J., and N. Limao. 1999. "Geographical Disadvanage: a Heckscher-Ohlin-von Thunen Model of Inernaional Specialisaion." Journal of Inernaional Economics 58:239-63. Walz, U. 1996. "Transpor Coss, Inermediae Goods, and Localized Growh." Regional Science and Urban Economics 26:671-95. Woo, W.T. 1998. "Chinese Economic Growh: Sources and Prospecs." In Michel Fouquin e Françoise Lemoine, eds., The Chinese Economy. London: Economica. 22
Annex 1. Map of China 23