ADB Economics Working Paper Series

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications
Minsoo Lee No. 297 | January 2012

ADB Economics Working Paper Series No. 297

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications

Minsoo Lee January 2012

Minsoo Lee is Senior Economist in the Macroeconomics and Finance Research Division, Economics and Research Department, Asian Development Bank. The author accepts responsibility for any errors in the paper.

Asian Development Bank 6 ADB Avenue, Mandaluyong City 1550 Metro Manila, Philippines www.adb.org/economics 2012 by Asian Development Bank January 2012 ISSN 1655-5252 Publication Stock No. WPS124550 The views expressed in this paper are those of the author(s) and do not QHFHVVDULO\ UHHFW WKH YLHZV RU SROLFLHV of the Asian Development Bank.

The ADB Economics Working Paper Series is a forum for stimulating discussion and eliciting feedback on ongoing and recently completed research and policy studies undertaken by the Asian Development Bank (ADB) staff, consultants, or resource persons. The series deals with key economic and development problems, particularly WKRVH IDFLQJ WKH $VLD DQG 3DFLF UHJLRQ DV ZHOO DV FRQFHSWXDO DQDO\WLFDO RU methodological issues relating to project/program economic analysis, and statistical data and measurement. The series aims to enhance the knowledge on Asias development DQG SROLF\ FKDOOHQJHV VWUHQJWKHQ DQDO\WLFDO ULJRU DQG TXDOLW\ RI $'%V FRXQWU\ SDUWQHUVKLS VWUDWHJLHV DQG LWV VXEUHJLRQDO DQG FRXQWU\ RSHUDWLRQV DQG LPSURYH WKH TXDOLW\ DQG availability of statistical data and development indicators for monitoring development effectiveness. 7KH $'% (FRQRPLFV :RUNLQJ 3DSHU 6HULHV LV D TXLFNGLVVHPLQDWLQJ LQIRUPDO SXEOLFDWLRQ ZKRVH WLWOHV FRXOG VXEVHTXHQWO\ EH UHYLVHG IRU SXEOLFDWLRQ DV DUWLFOHV LQ SURIHVVLRQDO journals or chapters in books. The series is maintained by the Economics and Research Department.

Contents
Abstract I. II. III. Introduction Theory and Institutional Background Models and Methodology A. B. IV. V. Data Empirical Results A. B. C. VI. VII. R&D-Intensive Firms and Different Type of Exporters Heterogeneity in Origin of Investment Funds Processing Traders versus Normal Traders Tax Competition Model Conditional Logit Model v 1 3 5 5 6 8 9 9 12 15 18 20 21

Policy Implications Conclusion

References

Abstract
During 20002006, nearly 4,0005,000 new foreign and domestic enterprises entered the export industry of the Peoples Republic of China (PRC) yearly, with geographic concentration along the coastal areas. This paper empirically analyzes the spatial determinants of new entrants by applying traditional comparative advantage analysis, new economic geography of agglomeration WKHRU\ DQG WD[ FRPSHWLWLRQ WKHRU\ E\ UHVRXUFH RZ PRGHO 7KH KHWHURJHQHLW\ RI UP RZQHUVKLS DQG WUDGH W\SH HQKDQFHV WKH XQGHUVWDQGLQJ RI LQYHVWPHQW ORFDWLRQ decisions. Processing traders differ substantially from nonprocessing traders in response to changes in wage cost, level of development in infrastructure, DQG SUR[LPLW\ WR WKH 3DFLF FRDVW 7KH SUHVHQFH RI IRUHLJQLQYHVWHG HQWHUSULVHV FUHDWHV D EHDFRQ HIIHFW RQ VXEVHTXHQW IRUHLJQ DQG GRPHVWLF LQYHVWRUV 7KH SDSHU DOVR QGV WKDW DJJORPHUDWLRQ SOD\V D JUHDWHU UROH IRU LQYHVWRUV VSHFLDOL]LQJ in the processing goods trade and highly skill-intensive or research and GHYHORSPHQWLQWHQVLYH SURGXFWLRQ %HQHWV IURP UHGXFWLRQ LQ WKH YDOXH DGGHG tax rate and corporate income tax rate play an evident role for foreign investors and Hong Kong, China and Taipei,China investors in processing trade sector. The PRCs policy makers intend to transform the internationalization process through attraction of foreign investment toward localization by reducing the processing good trading portion and by increasing the supply of domestically produced intermediate inputs for their exports. As the PRC moves up the highvalue supply chain, expands its domestic market, and exports skill-intensive goods, processing goods traders may consider new locations or business models.

I. Introduction
Ever since the Peoples Republic of China (PRC) opened up its economic door to the rest of the world in the late 1970s, two economic aspects have been the center of discussion by many academics, researchers, and policy makers in the world. They are the PRCs inward foreign direct investment (FDI) and its exports. Those two issues were reemphasized when the PRC joined the World Trade Organization in 2001. Each year, during 20002006, nearly 4,0005,000 new foreign and domestic enterprises entered into the PRCs export industry with geographic concentration along the coastal areas. The heterogeneity of new enterprises in ownership is apparent. In the nation as a ZKROH GXULQJ  IRUHLJQLQYHVWHG UPV DFFRXQW IRU   +RQJ .RQJ &KLQD DQG RU 7DLSHL&KLQDLQYHVWHG HQWHUSULVHV   DQG &KLQHVH SULYDWH FRPSDQLHV   7KH QXPEHU RI SURFHVVLQJ JRRGV H[SRUWHUV DPRQJ QHZ HQWUDQWV VWDQG DW DSSUR[LPDWHO\   3URFHVVLQJ JRRGV H[SRUW YDOXH LQ WKH 35& FRQVWLWXWHV DERXW  RI WRWDO H[SRUWV LQ  DQG KDV VWHDGLO\ GHFOLQHG IURP  IURP  WR  Traditional comparative advantage analysis is based on a level of regional development VXFK DV LQIUDVWUXFWXUH LQ KLJKZD\V UDLOZD\V DQG ZDWHUZD\V SURGXFWLRQ FRVWV DEXQGDQFH RI LQSXWV ODERU ODZV hukou KRXVHKROG UHJLVWUDWLRQ V\VWHP HGXFDWLRQ OHYHO KXPDQ FDSLWDO ZDJH OHYHO WUDQVSRUWDWLRQ FRVWV SUR[LPLW\ WR WKH FRDVW DFFHVV WR UDZ DQG LQWHUPHGLDWH PDWHULDOV DQG SURGXFWLRQ FRVWV 7KH QHZ HFRQRPLF JHRJUDSK\ WKHRU\ RI DJJORPHUDWLRQ SURYLGHV VLJQLFDQW H[SODQDWRU\ SRZHU WR WKH SUHVHQFH RI VLPLODU UPV VXSSOLHU UPV RU FRQVXPHUV LQ D SDUWLFXODU UHJLRQ ZKLFK FDXVHV WKLV ORFDWLRQ WR EHFRPH more attractive to future investors. Local government policy makers in the PRC strategically determine their tax rates to attract new foreign investments to their region for additional tax revenue and new HPSOR\PHQW ZKLFK FDQ EH GHVFULEHG E\ WKH WD[ FRPSHWLWLRQ ZLWK UHVRXUFH RZ PRGHO Policy makers provide preferential tax treatment and/or lower regulatory burdens and often implement special policies such as special economic zones and hi-tech zones. There are a number of policy variables, such as capturing overall levels of tax and regulatory burdens or special policies, designed to attract foreign investors. This paper discusses the PRCs inward FDI and exports by raising the topic of the spatial determinants of new foreign- and domestic-invested enterprises that enter the export LQGXVWU\ ([SODQDWRU\ IDFWRUV RI VSDWLDO GHWHUPLQDQWV RI )', LGHQWLHG LQ WKLV SDSHU DUH categorized according to three economic theories: traditional comparative advantage

2 | ADB Economics Working Paper Series No. 297

analysis, new economic geography of agglomeration effects, and tax competition theory E\ UHVRXUFH RZ PRGHO This paper provides rich empirical results by mapping and merging two separate data VRXUFHV LQGXVWU\ GDWD DW WKH UP OHYHO IURP WKH 35&V ,QGXVWULDO (QWHUSULVHV 'DWDEDVH SURYLGHG E\ WKH 1DWLRQDO %XUHDX RI 6WDWLVWLFV DQG UPOHYHO H[SRUWLPSRUW GDWD IURP WKH Export/Import Database provided by the PRC Customs). In particular, new enterprises DUH FDWHJRUL]HG EDVHG RQ WKHLU RZQHUVKLS IRUHLJQLQYHVWHG +RQJ .RQJ &KLQD DQGRU 7DLSHL&KLQDLQYHVWHG &KLQHVH SULYDWHRZQHG DQG VWDWHRZQHG DQG DUH GLVWLQJXLVKHG EHWZHHQ UPV VSHFLDOL]LQJ LQ SURFHVVLQJ WUDGH DQG QRUPDO WUDGH 5RGULN  $PLW\ DQG )UHXQG  )XUWKHUPRUH WKH SDSHU VHSDUDWHV UPV DFFRUGLQJ WR WKH HGXFDWLRQ OHYHO RI WKHLU HPSOR\HHV DQG SUR[\ UHVHDUFK DFWLYLW\ E\ VHOHFWLQJ UPV WKDW HPSOR\ ZRUNHUV ZLWK SRVWJUDGXDWH TXDOLFDWLRQV 7KLV SDSHU VKRZV WKDW WKHVH VXEVDPSOHV GLVSOD\ economically meaningful differences. )RUHLJQLQYHVWHG SURFHVVLQJ WUDGH UPV SXW PRUH ZHLJKW RQ SUR[LPLW\ WR WKH 3DFLF FRVW and increase in average wages than normal traders do, and the presence of foreigninvested enterprises creates a beacon effect in agglomeration to all other types of export enterprises. Investors from Hong Kong, China and Taipei,China are the most inclined WR DJJORPHUDWH ZLWK FRPSDQLHV WKDW DUH DOVR IURP +RQJ .RQJ &KLQD DQG 7DLSHL&KLQD DQG WHQG WR OHDVW RFN WRJHWKHU ZLWK IRUHLJQLQYHVWHG UPV DQG VWDWHRZQHG HQWHUSULVHV 3ULYDWH &KLQHVH LQYHVWRUV DUH JUHDWO\ LQXHQFHG E\ WKH SUHVHQFH RI VWDWHRZQHG LQFXPEHQW UPV Value-added tax rate that is strategically determined by local government policy makers LQXHQFHV WKH ORFDWLRQ FKRLFH RI IRUHLJQ LQYHVWRUV )RUHLJQLQYHVWHG SURFHVVLQJ JRRGV WUDGHUV DUH SDUWLFXODUO\ VHQVLWLYH WR WKLV SUDFWLFH DQG &KLQHVH SULYDWHLQYHVWHG UPV DUH the least sensitive. The method of the analysis in this paper suggests useful information for policy makers and market entrants. Marginal effects and predicted changes are calculated, induced either by a certain percent change in each explanatory variable in the empirical DQDO\VLV LQXHQFLQJ WD[ UDWHV DQG ODERU FRVWV RU E\ LQGXFLQJ PDUNHW HQWU\ E\ SLRQHHULQJ LQFXPEHQW UPV RQ VXEVHTXHQW PDUNHW HQWUDQWV LQ YDULRXV VXEJURXSV VRPH RI ZKLFK PD\ EH SDUWLFXODUO\ GHVLUDEOH WR SROLF\ PDNHU LQWHQWLRQV RQ DFTXLULQJ QHZ WHFKQRORJ\ IURP foreign-invested enterprises via technology and management knowledge spillover effects in their province.

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 3

II. Theory and Institutional Background

The new economic geography theory in the international trade theory framework introduced by Krugman (1991) may be the most relevant to the spatial determinants of )', ,Q PRQRSROLVWLF PDUNHW VWUXFWXUHV HFRQRPLHV RI VFDOH SXVK PDQXIDFWXULQJ UPV to locate in the region with larger demand while minimizing transportation costs. The HFRQRPLF UHJLRQV ZLWK WKH PRVW SURGXFWLRQ EHFRPH PRUH SURWDEOH DQG DWWUDFW HYHQ PRUH SURGXFWLRQ DFWLYLWLHV 7KH JHRJUDSKLF FKRLFH RI H[SRUWLQJ UPV LV EDVHG RQ WKH DJJORPHUDWLRQ DVVRFLDWHG ZLWK VLPLODU LQGXVWU\ W\SHV RI LQFXPEHQW UPV DQG LQWHUDFWLRQ RI economies of scale with transportation costs (proximity to coast). Geographic productionFRQFHQWUDWLRQ ORZHUV WKH SUREDELOLW\ RI ODERU VKRUWDJHV ,QFXPEHQW UPV FDQ VSLOO RYHU informational, technology, and knowledge externally. The PRCs dynamic policy changes seem to follow Alfred Marshalls concept regarding industry localization: internationalization toward localization or processing trades to QRUPDO WUDGHV &RQFHQWUDWLRQ RI IRUHLJQLQYHVWHG UPV LQ D ORFDWLRQ JHQHUDWHV ODERU PDUNHWV IRU D VSHFLF LQGXVWU\ /RFDO ODERU PDUNHWV VXSSO\ ORZVNLOOHG ODERU DQG IRUHLJQ SURFHVVLQJ WUDGHUV LPSRUW VRSKLVWLFDWHG LQSXWV KRZHYHU ORFDO SURGXFHUV ZLOO UHSODFH intermediate imports and instead start producing sophisticated inputs in the near future. $JJORPHUDWHG IRUHLJQ UPV ZLOO EHQHW IURP LQIRUPDWLRQ DQG NQRZOHGJH VSLOORYHUV ZKLOH GRPHVWLF UPV FDQ EHQHW IURP WHFKQRORJ\ DQG SRVVLEOH NQRZOHGJH VSLOORYHUV IURP foreign-invested enterprises. &DUOWRQV  DUWLFOH DQ H[FHSWLRQDOO\ VLJQLFDQW SDSHU FRQFHUQLQJ HFRQRPLF JHRJUDSK\ HPSKDVL]HV ORFDWLRQ DQG HPSOR\PHQW RI QHZ UPV XWLOL]LQJ DQ HFRQRPHWULF model with discrete and continuous endogenous variables. Carlton concludes that taxes DQG VWDWH LQFHQWLYH SURJUDPV GR QRW DIIHFW D UPV GHFLVLRQ Amiti and Javorcik (2008) analyze new economic geography of agglomeration effects in PRC. They use aggregated provincial-level data to generate measures of intermediate goods and consumer market access using recent Chinese inputoutput tables. Access WR VXSSOLHUV DQG PDUNHWV LV WKH PRVW LQXHQWLDO VSDWLDO GHWHUPLQDQWV RI D UPV ORFDWLRQ choice. Hu and Owen (2005), however, contend that there is no evidence of strong agglomeration from foreign investors in the PRC but conclude that comparative DGYDQWDJH DQG JRYHUQPHQW SROLF\ DUH PRUH LQXHQWLDO IDFWRUV IRU D UPV VSDWLDO GHFLVLRQ DV H[SODQDWLRQV RI WKH HPHUJLQJ SDWWHUQ RI UP ORFDWLRQV 2YHUVHDV &KLQHVH LQYHVWRUV located along the coast show sensitivity to policy and cost differentials compared to OECD investors. In inland provinces, on the other hand, there seems to be no difference between the two groups of investors. Boermans et al. (2009) use factor analysis with aggregated data and emphasize the major role of institutions, labor costs, and market access.

4 | ADB Economics Working Paper Series No. 297

A seminal paper by Head and Ries (1996) formulates a theoretical monopolistic competitive model of self-reinforcing FDI. The paper explains that open door policies made inward FDI on a large scale to the PRC feasible and their empirical results, comprised of 931 foreign ventures that entered between 1984 and 1991 in 54 cities, LOOXVWUDWH FRQVLGHUDEOH DJJORPHUDWLRQ HIIHFWV WKDW PDJQLHG WKH DGYDQWDJH FRQIHUUHG E\ the privileged status of some cities such as special economic zones and high-tech zones. Parallel literature explores decisions in particular industries in the PRC by investors from different countries. Wakasugi (2005) demonstrates that location choice of Japanese HOHFWURQLFV DIOLDWHV LQ WKH 35& PDLQO\ GHSHQG RQ WKH OHYHO RI KXPDQ UHVRXUFHV &KDQJ and Park (2005) use data for Korean investors in the PRC, indicating an importance of network externality. Belderbos and Carree (2002) examine Japanese manufacturers ORFDWLRQV LQ WKH 35& DQG VKRZ VLJQLFDQW HFRQRPLF JHRJUDSK\ RI DJJORPHUDWLRQ HIIHFWV &KHQV  HPSLULFDO DQDO\VLV RQ 7DLSHL&KLQDV JUHHQHOG LQYHVWPHQW LQ WKH 35& HPSOR\HG D QHVWHGORJLW PRGHO ZLWK UPOHYHO GDWD ZKLFK LQGLFDWHG WKDW WKH DJJORPHUDWLRQ HIIHFW IURP WKH LQFXPEHQW UPV KDV EHFRPH PRUH DSSDUHQW RYHU WLPH EXW its relative strength varies across sectors. ([SODQDWRU\ YDULDEOHV LGHQWLHG E\ WKH DERYH PHQWLRQHG VWXGLHV W\SLFDOO\ IDOO LQWR WKUHH FDWHJRULHV PHDVXUHV RI D UHJLRQDO OHYHO RI GHYHORSPHQW +X DQG 2ZHQ  QHZ economic geography theories (Amiti and Javorcik 2008) of agglomeration effects with the SUHVHQFH RI VLPLODU UPV LQWHUPHGLDWH JRRGV VXSSOLHUV DQG FRQVXPHUV LQ D SDUWLFXODU UHJLRQ PDNLQJ D ORFDWLRQ PRUH DWWUDFWLYH WR IXWXUH LQYHVWRUV DQG SROLF\ YDULDEOHV FDSWXULQJ overall levels of tax and regulatory burden or special policies designed to attract foreign investors (Head and Ries 1996). Many studies in new economic geography analysis have omitted some obvious JRYHUQPHQW SROLF\ YDULDEOHVD PHDVXUH RI D UPV WD[ EXUGHQ RU RWKHU SUHIHUHQWLDO treatment variables. Cheng and Kwan (2000) divide tax reduction in the PRC into four levels as to provincial VFRSH ZLWK WKH FRQFOXVLRQ WKDW WD[ UHGXFWLRQ LV WKH PRVW LPSRUWDQW IDFWRU RI )', LQRZ There have been a large number of government regulations on tax reductions at the central and local government level since 1984. These tax breaks and tax reduction SROLFLHV KDYH PDLQO\ WZR WDUJHWV 7KH UVW LV FRUSRUDWLRQ LQFRPH WD[ 5HJXODWLRQV LQ WKH V RIIHU  FRUSRUDWLRQ LQFRPH WD[ UDWH IRU IRUHLJQLQYHVWHG FRPSDQLHV LQ YDULRXV HFRQRPLF DQG WHFKQRORJ\ GHYHORSPHQW ]RQHV  FRUSRUDWLRQ LQFRPH WD[ UDWH IRU FRPSDQLHV LQ FRDVWDO HFRQRPLF RSHQ ]RQHV ZLWK H[SRUWV FRQVWLWXWLQJ PRUH WKDQ  RI WRWDO RXWSXW RU DQ H[HPSWLRQ LQ FRUSRUDWLRQ LQFRPH WD[ LQ WKH UVW  \HDUV RI LWV RSHQLQJ and reduction by half of income tax in the following 5 years. The second is value-added tax, which is the main resource of tax revenue. There are regulations offering various UDWHV RI YDOXHDGGHG WD[ UDWHV IRU UPV LQ WKH SURFHVVLQJ LQGXVWU\ DGPLWWHG E\ ORFDO authority. Value-added tax revenue in the PRC is shared by central and local government

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 5

ZLWK UDWLRV RI  DQG   UHVSHFWLYHO\ 7KHUH LV DOVR VDOHV WD[ DQG DQ DGGLWLRQDO accounting item including sales tax, property tax, goods and services tax, resource tax, land value increment tax, urban maintenance and construction tax, and surcharge for education and local education expenses. New legislation in 1994 made local governments more independent units motivated to PD[LPL]H WKHLU SROLWLFDO DQG HFRQRPLF EHQHWV DV WKH\ FRPSHWH WR DWWUDFW JUDQWV DQG valuable factors to their regions (Qian and Roland 1998). This competition among local DXWKRULWLHV EHFRPHV D IRFXV RI WKHRUHWLFDO ZRUN RZLQJ WKUHH OLWHUDWXUHV WD[ FRPSHWLWLRQ literature, welfare competition literature, and spillover effects literature. With little power on the legislation of taxation, local governments in the PRC practice tax break policies in order to reduce tax for grants, expand tax-favored zones such as economic and technological development zones or high-tech industrial zones, or lower VWDQGDUGV RI WD[ EUHDN TXDOLFDWLRQV IRU FRUSRUDWLRQV ZKLFK PDNHV WD[ FRPSHWLWLRQ QR longer observable or measurable by tax rates announced in legislative documents. Tax reduction by local governments should be conducted mostly by using tax break policies, which vary over regions and industries. The most important motivation for tax reduction is to attract FDI, and this is mentioned in a multitude of research papers. Brueckner (2002) illustrates this tax competition SKHQRPHQRQ E\ WKH UHVRXUFH RZ PRGHO :LOVRQV  VXUYH\ SDSHU SURYLGHV D ZLGH range of ideas about the tax competition model.

III. Models and Methodology

A. Tax Competition Model

Local governments levy taxes on a mobile tax base (foreign and domestic investments) by choosing their tax rates strategically with particular attention to other local government WD[ UDWHV 7D[ FRPSHWLWLRQ PRGHOV DUH PRVWO\ ZLWKLQ WKH IUDPHZRUN RI WKH UHVRXUFH RZ model summarized by Brueckner (2002) in the review of government interaction. This model is followed by this paper, which uses an empirical model according to tax reaction function. The general form of tax reaction function can be generated from various tax competition theoretical models, as shown in the seminal work by Brueckner and Saavedra (2001). Each local government chooses its tax rate tj and considers the amount of foreign investments (resource) within its borders. Distribution of foreign investments in provinces depends on t WKHUHIRUH jth local government decision on tax rate is indirectly affected by

6 | ADB Economics Working Paper Series No. 297

tj (vector of ts for the n-1 other local governments). The local government js objective function is V (t j , s j ; X j ) , where sj LV DQ DPRXQW RI LQRZ RI IRUHLJQ LQYHVWPHQW LQWR jth province and Xj is the vector of province js characteristics. Distribution of sj is affected by t. s j = H (t j , t j ; X j )

7KHUHIRUH WKH UHGXFHG IRUP RI WKH UHVRXUFHRZ PRGHO FDQ EH GHULYHG DV V t j , H (t j , t j ; X j ) ; X j V (t j , t j ; X j ) Province jV RSWLPDO WD[ UDWH DV D UHDFWLRQ IXQFWLRQ IURP WKH UVW RUGHU FRQGLWLRQ WDNHV WKH form of t j = R (t j ; X j ) .

B.

Conditional Logit Model

In connection with the comparative advantage theory and new economic geography of DJJORPHUDWLRQ DQDO\VLV WKLV VWXG\ DGRSWV WKH PRGLHG DGGLWLYH UDQGRP XWLOLW\ PRGHO E\ VXEVWLWXWLQJ H[SHFWHG SURWV 3 for utility U. Suppose an investor i chose one province among j   DOWHUQDWLYHV WKHQ H[SHFWHG SURWV IRU DQ LQYHVWRU i of choosing alternative j y i = j is given by

E ij = Vij + H ij = V CAij , NEGij , t j + H ij , where Vij LV WKH GHWHUPLQLVWLF SDUW RI H[SHFWHG SURWV DQG H ij represents a random part (McFadden 1974). it can be expected that investors choose one province among j = 1,, J so as to maximize ij . The probability of choosing any one alternative j is HTXLYDOHQW WR WKDW WKH DOWHUQDWLYH \LHOGV JUHDWHU H[SHFWHG SURWV WKDQ DOO RWKHU DOWHUQDWLYH choices: pij = Pr yi = j = Pr ij ik , k j = Pr H ik H ij Vij Vik , k j In order to solve for these choice probabilities, an appropriate joint distribution of the UDQGRP SDUWV PXVW EH VSHFLHG 0F)DGGHQ  SURYHV WKDW WKH IROORZLQJ H[SUHVVLRQ F H i 1,, H iJ = exp G e H i 1 ,, e H iJ

( )

( (

))

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 7

is the cumulative distribution function of a multivariate extreme value distribution for any function G WKDW LV QRQQHJDWLYH KRPRJHQHRXV RI GHJUHH RQH DQG IXOOOV D QXPEHU RI additional regularity conditions (McFadden 1978). Different choices of G yield different models. The choice probabilities can be calculated given that the random parts of H[SHFWHG SURWV DUH GLVWULEXWHG DFFRUGLQJ WR IXQFWLRQ F (McFadden 1978): pij = Pr y i = j : investor is choice = j = =
+

) F (H ,V
j

V2 + H ,,V1 VJ + H d H

G1 e H , e H V1 +V2 ,, e H V1 +VJ exp G e H , e H V1 +V2 ,, e H V1 +VJ

) ( (

)) dH

ZKHUH WKH VXEVFULSWV WR ) DQG * GHQRWH WKH UVW GHULYDWLYH 6LQFH * LV KRPRJHQHRXV RI degree 1, its derivative G1 is homogeneous of degree zero. Therefore

pij = Pr y i = j =

) e

G j eV1 , eV2 ,, eVJ exp e H e

) (

V j

G eV1 , eV2 ,, eVJ

)) dH

=e

Vj

G1 eV1 , eV2 ,, eVJ

V1

( G (e
(

, eV2 ,, eVJ

) ) , , j {1,, J}.
e
Vij

Assume that the errors H j are iid with Weibull distribution, then pij = Pr y i = j =

eVi 1 + eVi 2 +  + eViJ

6LQFH DOO RI DOWHUQDWLYHVSHFLF H[SODQDWRU\ YDULDEOHV YDU\ DFURVV DOWHUQDWLYHV FRQGLWLRQDO logit model is most suitable for this case. pij = Pr y i = j =

exp ECACAij + E NEG NEGij + E t t ij

J k =1

exp ECACAik + E NEG NEGik + E t t ik

The standard maximum likelihood estimation can be used if the choice probabilities DERYH DUH FRUUHFWO\ VSHFLHG 6LQFH WKH PRGHO LV QRQOLQHDU LQWHUSUHWDWLRQ RI WKH HVWLPDWHG FRHIFLHQWV LV QRW VWUDLJKWIRUZDUG 7R JDXJH WKH PDJQLWXGH RI HVWLPDWHG HIIHFWV WKH marginal effects of changes in explanatory variables on choice probabilities have to be FDOFXODWHG %\ PXOWLSO\LQJ ZLWK WKH FRHIFLHQW HVWLPDWH DQG GLYLGLQJ E\ WKH SUREDELOLW\ these marginal effects can be converted into elasticities. Cameron and Trivedi (2005)

8 | ADB Economics Working Paper Series No. 297

illustrate that the closed form expression of the marginal effects in the conditional logit model is: pij xik

= pij i ijk pik E ,

i where ijk is an indicator variable that for an investor i takes on the value 1 LI M N DQG  RWKHUZLVH ,W LV FOHDU IURP WKH DERYH IRUPXOD WKDW WKLV GHQLWLRQ LPSOLHV WKDW IRU H[DPSOH D SRVLWLYH UHJUHVVLRQ FRHIFLHQW PHDQV WKDW D PDUJLQDO LQFUHDVH LQ WKH x covariate ik will lead to a rise in the predicted probability of option k being chosen, but to a fall in the predicted probability of any other option being chosen.

IV. Data
Three major data sources are utilized for empirical analysis. The National Bureau of 6WDWLVWLFV SURYLGHV DQQXDO UPOHYHO LQGXVWU\ GDWD VHWV &KLQHVH ,QGXVWULDO (QWHUSULVHV 'DWDEDVH FRQWDLQLQJ LQIRUPDWLRQ RQ DOO UPV RSHUDWLQJ LQ PDQXIDFWXULQJ VHFWRUV ZKRVH DQQXDO UHYHQXH LV JUHDWHU WKDQ  PLOOLRQ \XDQ $Q H[WHQVLYH DUUD\ RI UPOHYHO YDULDEOHV LV LQFOXGHG LQ WKLV GDWD VHW VXFK DV D UPV RZQHUVKLS FRPSRVLWLRQ RI HPSOR\PHQW FRUSRUDWH LQFRPH WD[ DPRXQW YDOXH DGGHG WD[ DPRXQW UPV ORFDWLRQ HWF ,QGXVWU\ GDWD in 2004 contains the established date of each company. New economic geography of DJJORPHUDWLRQ HIIHFWVLQ RWKHU ZRUGV DQ LQXHQFH RI WKH SUHVHQFH RI VLPLODU UPV LQ FHUWDLQ ORFDWLRQV RQ QHZ HQWUDQWV ORFDWLRQ GHFLVLRQVFDQ EH WHVWHG E\ SURSHUO\ GHQLQJ DQ LQFXPEHQF\ $OO H[SRUWLQJ UPV WKDW HQWHUHG WKH PDUNHW SULRU WR  ZLOO EH FRXQWHG as incumbents from the perspective of a new market entrant in 2004. Incumbency statistics are constructed at the 4-digit Harmonized System (HS)-code level associated ZLWK WKH RZQHUVKLS RI UPV IRUHLJQLQYHVWHG +RQJ .RQJ &KLQD7DLSHL&KLQD +7 RZQHG VWDWHRZQHG DQG &KLQHVH SULYDWHRZQHG 7KH RZQHUVKLS VWDWXV LV GHQHG DV D PDMRULW\ share of capital. The [Peoples Republic of] China Customs records monthly export/import data at the transaction level, including sector information in the form of 8-digit HS-codes, among other variables like unit value, type of enterprises, commodity code, etc. This study maps/ merges Industry data and Customs data by matching the company names, and then the VXEVDPSOH RI H[SRUWLQJ UPV IURP WKH PHUJHG GDWD VHW DUH UHVHUYHG IRU WKH DQDO\VLV (DFK UPV H[SRUW WUDQVDFWLRQ GRFXPHQWHG LQ WKH H[SRUWLPSRUW GDWD VHW KHOSV WR LGHQWLI\ LWV PDLQ HOG RI EXVLQHVV HVWDEOLVKLQJ LWV VHFWRU LGHQWLW\ 7KH GLVDJJUHJDWLRQ OHYHO LV at 4-digit HS-codes and all exports of products in the same 4-digit HS-code group are WUHDWHG DV HTXLYDOHQW

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 9

7KH >3HRSOHV 5HSXEOLF RI@ &KLQD &XVWRPV 6WDWLVWLFV FODVVLHV HYHU\ WUDQVDFWLRQ RI RZ of goods based on two trade types: processing and normal trade. Processing trade insinuates a low-value added step in the supply chain of a sophisticated product. Hightechnology intermediate imports are assembled by low-skilled local workers to produce VRSKLVWLFDWHG QDO H[SRUW JRRGV PDLQO\ VKLSSHG WR DGYDQFHG HFRQRPLHV 1HYHUWKHOHVV the assembling work performed by local workers contributes limited prospects of skill upgrading or technology/knowledge spillover for the PRC to move up the high value supply chain, expand its domestic market, and export skill-intensive goods. Rodrik (2006) points out that the behavior of processing traders is therefore both of scholarly interest and relevant for policy makers. $ GLIFXOW FKDOOHQJH LV WR FRQVWUXFW D PHDVXUH RI DFFHVV WR LQWHUPHGLDWH JRRG VXSSOLHUV due to the lack of reliable information. Nevertheless, processing traders yield useful insights on this aspect. Processing traders are largely autonomous from local intermediate JRRGV SURGXFHUV RWKHU WKDQ ODERU WKHUHIRUH RQH FDQ VXFFHVVIXOO\ DEVWUDFW IURP WKH LVVXH of access to intermediate goods suppliers by analyzing processing traders separately from nonprocessing traders. Normal traders might have generated a strong network of intermediate goods suppliers, since incumbent enterprises presumably made location their consideration in choosing intermediate goods suppliers. The Annual Statistical Yearbook of [the Peoples Republic of] China contains extensive control variables at the province level. A range of variables at the province level is collected to model comparative advantage analysis of a market entrants location choices. ,Q DGGLWLRQ D URXJK GLVWDQFH EHWZHHQ D SURYLQFHV PDLQ FLWLHV DQG WKH 3DFLF FRDVW LV measured.

V. Empirical Results
A. R&D-Intensive Firms and Different Type of Exporters

Firms in the high-skill-intensive production industry that employ workers with postgraduate GHJUHHV DUH XVHG DV D SUR[\ IRU 5 'LQWHQVLYH UP DFWLYLW\ ZLWK UHVSHFW WR WHFKQRORJ\ knowledge upgrading and spillovers. 0RGHO  LQ 7DEOH  SURYLGHV WKH EDVHOLQH VSHFLFDWLRQ FRPSULVLQJ WKH ZKROH VDPSOHV DQG 0RGHO  H[KLELWV UHVXOWV IRU 5 'LQWHQVLYH UPV $V RSSRVHG WR WKH RYHUDOO HQWHUSULVHV 5 'LQWHQVLYH UPV SXW PRUH YDOXH RQ WKH FROOHJH JUDGXDWH SRSXODWLRQ DQG DUH VHQVLWLYH WR ERWK YDOXHDGGHG DQG FRUSRUDWH WD[ UDWHV ,QFXPEHQW 5 'LQWHQVLYH UPV attract new, similar types of companies through the agglomeration effect.

10 | ADB Economics Working Paper Series No. 297

Table 1: Conditional Logit Model: All Firms and R&D Intensive Firms
Model (1): All Firms Variables wage_manufacturing GDP per capita waterway highway high school pop college pop proximity value added tax rate corp income tax rate incumbent firms Sample size LL Chi2 Pseudo R2 Estimates Marginal Std. dev. Predicted Effects Change* 0.00 0.00 3016 3.02 [22.617] [27.590] 0.00 0.00 4910 1.47 [21.764] [16.826] 6.53 0.50 0.07 3.57 [51.405] [38.069] 0.97 0.07 0.21 1.56 [9.630] [9.076] 7.78 0.60 0.04 2.6 [12.188] [15.387] 0.59 0.05 0.03 0.15 [0.677] [0.666] 0.00 0.00 686 18.93 [57.455] [41.952] 1.18 0.09 0.07 0.62 [4.784] [4.939] 1.65 0.13 0.02 0.29 [3.202] [3.175] 0.01 0.00 65 3.58 [62.528] [31.150] 495673 41282 33203 0.29 Model (2): R&D-Intensive Firms Estimates Marginal Std. dev. Predicted Effects Change* 0.00 0.00 3008 3.61 [7.283] [8.512] 0.00 0.00 4777 0.96 [5.012] [4.656] 6.15 0.79 0.07 5.65 [20.745] [22.473] 1.30 0.17 0.21 3.52 [5.398] [5.063] 8.35 1.08 0.04 4.66 [6.547] [8.074] 7.55 0.97 0.03 3.26 [4.789] [5.859] 0.00 0.00 686 18.11 [17.935] [21.379] 0.73 0.10 0.07 0.65 [1.508] [1.533] 3.77 0.49 0.02 1.1 [3.395] [3.295] 0.01 0.00 51 5.22 [26.089] [16.116] 89217 7988 4868 0.23

* Predicted changes (in %) induced by a 1 standard deviation increase in the explanatory variable. Note: Figures in the square bracket are z-statistics. Source: Author's calculations.

0RGHOV  DQG  LQ 7DEOH  VKRZ DQDO\VHV IURP WKH KHWHURJHQHLW\ RI UPV LQ WKHLU trade type, according to whether they predominantly focus on processing trade or normal trade. Processing trade has been at the center of debates about the sophistication level of the PRCs exports (Rodrik 2006, Wang and Wei 2008). Processing trade is largely dependent on a limited range of local inputs, mainly labor. Processing traders import sophisticated intermediate goods, and only assembly and related low value-added production processing take place locally. Therefore, processing traders are relatively less dependent on intermediate goods supplier networks, while they are particularly sensitive to factors like labor costs, value added tax rate, regional development in infrastructure, DQG SUR[LPLW\ WR WKH 3DFLF FRDVW

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 11

Table 2: Conditional Logit Model: Processing Traders versus Normal Traders

Variables Model (3): Processing Traders Model (4): Normal Traders Estimates Marginal Std. dev. Predicted Effects Change* 0.00 0.00 3016 2.71 [24.052] [24.794] 0 0 4886 0.98 [13.511] [10.665] 7.55 0.40 0.07 2.87 [50.721] [26.129] 0.12 0.01 0.21 0.13 [1.030] [1.027] 12.08 0.64 0.04 2.8 [16.203] [21.596] 10.09 0.54 0.03 1.81 [10.476] [14.975] 0.00 0.00 686 13.17 [51.245] [27.258] 0.67 0.04 0.07 0.24 [2.261] [2.317] 1.23 0.07 0.02 0.15 [1.981] [1.968] 0.01 0.00 60 2.36 [50.676] [21.206] 341665 29534 20752 0.26 Estimates Marginal Std. dev. Predicted Effects Change* wage_manufacturing 0.00 0.00 3015 2.11 [5.329] [5.820] GDP per capita 0 0 4963 2.98 [17.388] [15.144] waterway 4.99 0.56 0.07 3.96 [19.295] [23.580] highway 3.78 0.42 0.21 8.88 [18.170] [15.932] high school pop 3.82 0.43 0.04 1.84 [3.043] [2.897] college pop 33.63 3.75 0.03 12.52 [17.142] [12.808] proximity 0.00 0.00 686 30.32 [26.313] [55.907] value added tax rate 3.61 0.40 0.07 2.76 [7.601] [7.883] corp income tax rate 2.32 0.26 0.02 0.58 [2.418] [2.395] incumbent firms 0.01 0.00 75 4.44 [28.221] [25.095] Sample size 154008 LL 11207 Chi2 13531 Pseudo R2 0.38

* Predicted changes (in %) induced by a 1 standard deviation increase in the explanatory variable. Note: Figures in the square bracket are z-statistics. Source: Author's calculations.

Comparing results between processing traders and normal traders supports intuition. From the regression results, there is indeed a difference in agglomerative effects. A new processing trader tends to select a location where other processing traders are operating. In fact, the predicted response to a 1 standard deviation increase in the number of LQFXPEHQWV LQ WHUPV RI FKRLFH SUREDELOLWLHV LV LQ WKH FDVH RI SURFHVVLQJ WUDGHUV   URXJKO\  SHUFHQWDJH SRLQWV DERYH WKH UHVSRQVH RI QRUPDO WUDGHUV   1RUPDO WUDGHUV LQ WKHLU ORFDWLRQ FKRLFHV DUH XQOLNHO\ LQXHQFHG E\ D UHGXFWLRQ LQ HLWKHU YDOXH added tax rate or corporate income tax rate while processing traders seriously consider WKH EHQHW IURP YDOXH DGGHG WD[ UDWH UHGXFWLRQ 7KH VDPH KROGV IRU SURFHVVLQJ WUDGHUV sensitivity to distance from the coast. A 1 standard deviation increase in the distance IURP WKH 3DFLF FRDVW LV DVVRFLDWHG ZLWK D PRUH WKDQ  UHGXFWLRQ LQ FKRLFH SUREDELOLW\ ZKHUHDV WKH UHGXFWLRQ IRU QRQSURFHVVLQJ WUDGLQJ UPV LV DERXW   The obtained results for the infrastructure variables show that there is a large effect of well-developed road infrastructure on the location decisions of processing traders. $  VWDQGDUG GHYLDWLRQ ULVH LQ URDGV SHU DUHD UDLVHV WKH FKRLFH SUREDELOLW\ E\   ZKHUHDV WKH HIIHFW LV LQVLJQLFDQW IRU QRUPDO WUDGHUV ZKLFK FRQIRUPV WR H[SHFWDWLRQV WKDW

12 | ADB Economics Working Paper Series No. 297

processing traders are particularly sensitive to transportation costs, and complements WKH ODUJH FRHIFLHQW HVWLPDWH IRU WKH GLVWDQFH IURP WKH FRDVW $W WKH VDPH WLPH KRZHYHU a comparable estimate is obtained for the waterway density of a provincea rise in waterway density has an effect of roughly the same magnitude for both processing and normal traders. Overall, it is noticeable that the model seems to do a better job at predicting the choices RI ORZYDOXH DGGHG SURFHVVLQJ UPV GLVSOD\LQJ D PXFK EHWWHU RYHUDOO PRGHO W ZLWK D SVHXGR5VTXDUHG RI  FRPSDUHG WR RWKHU VDPSOHV +LJK YDOXHDGGHG UPV PLJKW EH more dependent on highly specialized input factors (say, college graduates of a certain subject, or intellectual property rights), and therefore their choice propensities may yield predictions less accurate by broad provincial level aggregate variables.

B.

Heterogeneity in Origin of Investment Funds

In order to shed more light on the heterogeneous agglomerative processes in the PRCs H[SRUW LQGXVWU\ WKH LQFXPEHQW UPV DUH GLYLGHG E\ WKHLU RZQHUVKLS RULJLQ IRUHLJQLQYHVWHG )1 +RQJ .RQJ &KLQD7DLSHL&KLQD LQYHVWHG +7 VWDWHRZQHG 67 DQG &KLQHVH SULYDWHRZQHG &3 LQFXPEHQW UPV )XUWKHU DQDO\VLV ZLWK IRXU VXEJURXSHG LQYHVWRUV EDVHG RQ WKHLU RZQHUVKLS DOORZV RQH WR H[DPLQH ZKLFK UPV OLNHO\ SOD\ D VLJQLFDQW DJJORPHUDWLYH OHDGLQJ UROH IRU VXEVHTXHQW UPV LQ WKH VDPH RZQHUVKLS RULJLQ DQG different origin to their location decision. For example, foreign investors might follow other IRUHLJQ UPV RU WKH SUHVHQFH RI VWDWHRZQHG UPV PLJKW VKRZ D VWURQJ DJJORPHUDWLYH effect to Chinese private investors. Table 3 introduces separate incumbent counts depending on investor origin. For the IRUHLJQLQYHVWHG VDPSOH  ZDJH DQG SUR[LPLW\ YDULDEOHV DUH WKH PRVW LQXHQWLDO IDFWRUV WR QHZ HQWUDQWV IROORZHG E\ OHYHO RI LQIUDVWUXFWXUH GHYHORSPHQW %HQHWV IURP the reduction in value added tax rate are also taken critically by foreign investors while an incentive from corporative income tax provided by local policy makers is considered somewhat to a lesser extent.

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 13

Table 3: Conditional Logit Model: Foreign Firms and Hong Kong, China/Taipei,China Firms
Variables Model (5): Foreign-Owned Estimates Marginal Std. dev. Predicted Effects Change* 0.00 0.00 3037 3.34 [21.608] [15.832] 0.00 0.00 5057 1.01 [13.481] [8.602] 7.17 0.31 0.07 2.2 [26.881] [16.281] 1.57 0.07 0.21 1.43 [8.261] [6.643] 11.17 0.48 0.04 2.09 [8.853] [11.842] 9.71 0.42 0.03 1.41 [6.216] [9.015] 0.00 0.00 686 14.61 [32.301] [15.715] 3.06 0.13 0.07 0.9 [6.565] [6.864] 2.87 0.12 0.02 0.28 [3.662] [3.751] 0.01 0.00 26 0.66 [14.244] [9.124] 0.02 0.00 3 0.29 [4.038] [3.878] 0.00 0.00 33 0.12 [1.990] [1.859] 0.02 0.00 20 1.82 [26.019] [10.406] 143392 11410 10652 0.32 Model (6): Hong Kong, China/ Taipei,China-Owned Estimates Marginal Std. dev. Predicted Effects Change* 0.00 0.00 3005 9.02 [13.435] [9.835] 0.00 0.00 4833 2.9 [13.783] [11.800] 7.38 0.97 0.07 6.88 [22.978] [21.786] 0.31 0.04 0.21 0.84 [1.165] [1.168] 7.15 0.94 0.04 4.05 [5.398] [4.836] 50.36 6.61 0.03 22 [20.445] [13.251] 0.00 0.00 686 34.02 [20.999] [53.876] 1.41 0.19 0.07 1.27 [1.925] [1.946] 3.99 0.52 0.02 1.18 [2.138] [2.153] 0.00 0.00 27 0.97 [5.345] [5.115] 0.00 0.00 3 0.07 [0.374] [0.374] 0.01 0.00 35 2.36 [13.869] [13.375] 0.01 0.00 21 1.74 [7.192] [6.955] 126650 8330 12901 0.44

wage_manufacturing GDP per capita waterway highway high school pop college pop proximity value added tax rate corp income tax rate Incumbent_CP Incumbent_ST Incumbent_HT Incumbent_FN Sample size LL Chi2 Pseudo R2

* Predicted changes (in %) induced by a 1 standard deviation increase in the explanatory variable. Note: Figures in the square bracket are z-statistics. Source: Author's calculations.

$OO IRXU +6 OHYHO LQFXPEHQW VWDWLVWLFV DUH VLJQLFDQW DQG WKH ODUJHVW DJJORPHUDWLRQ HIIHFW LV DVVRFLDWHG ZLWK IRUHLJQRZQHG LQFXPEHQW UPV 7ZHQW\ DGGLWLRQDO IRUHLJQRZQHG LQFXPEHQWV LQ D SURYLQFH ZRXOG LQFUHDVH WKH FKRLFH SUREDELOLW\ E\ DERXW   DQG WKH HIIHFW RI &KLQHVH SULYDWH UPV LV FRPSDUDEOH +RZHYHU WKH DJJORPHUDWLRQ HIIHFW RI +7 LQYHVWHG UPV LV QHJDWLYH LQ RWKHU ZRUGV IRUHLJQ LQYHVWRUV DUH OHVV OLNHO\ WR FKRRVH D location where HT-invested companies are operating. New foreign investors seem to consider enterprises from Hong Kong, China and Taipei,China as competitors while they LQWHUSUHW WKH SUHVHQFH RI RWKHU IRUHLJQ UPV DV SRWHQWLDO FRUSRUDWLYH EXVLQHVV SDUWQHUV RU WKH\ UHFRJQL]H WKH YDOXDEOH H[SHULHQFH RI LQFXPEHQW IRUHLJQ UPV RSHUDWLQJ LQ D VLPLODU industry sector in the PRC.

14 | ADB Economics Working Paper Series No. 297

Table 4: Conditional Logit Model: Chinese Private Firms

Variables Model (7): Chinese Private Firms Estimates wage_manufacturing GDP per capita waterway highway high school pop college pop proximity value added tax rate corp income tax rate Incumbent_CP Incumbent_ST Incumbent_HT Incumbent_FN Sample size LL Chi2 Pseudo R2 0.00 [11.434] 0.00 [13.114] 3.38 [14.531] 0.56 [3.507] 7.64 [7.324] 0.65 [0.498] 0.00 [31.890] 0.26 [0.713] 1.69 [2.044] 0.02 [43.597] 0.05 [11.628] 0.00 [3.235] 0.01 [13.416] Marginal Std. dev. Effects 0.00 3009 [14.028] 0.00 4874 [10.252] 0.31 0.07 [17.384] 0.05 0.21 [3.391] 0.71 0.04 [9.234] 0.06 0.03 [0.508] 0.00 686 [23.249] 0.02 0.07 [0.719] 0.16 0.02 [2.006] 0.00 26 [16.823] 0.01 3 [10.001] 0.00 26 [3.389] 0.00 15 [9.910] 208699 17378 14018 0.29 Predicted Change* 3.01 1.95 2.23 1.08 3.07 0.2 16.33 0.16 0.35 4.06 1.44 0.35 1.79

* Predicted changes (in %) induced by a 1 standard deviation increase in the explanatory variable. Note: Figures in the square bracket are z-statistics. Source: Author's calculations.

In Model (6), HT investors show a preference in their entry decision for a province with a large high school graduate population, but a strong restriction for a province with a large college graduate population. Most of HT investors in the PRC mainly seek low-skilled laborers while some foreign investors may need intermediate-level managers with college degrees. The obtained result for the wage variable, however, is somewhat puzzling with a positive sign. On one hand, provincial individual income level greatly affects the location GHFLVLRQV RI LQYHVWRUV IURP +RQJ .RQJ &KLQD DQG 7DLSHL&KLQD %HQHWV IURP WKH reduction in both value-added tax rate and corporate income tax rate are taken intensely by investors from Hong Kong, China and Taipei,China. The results also reveal a much stronger pattern of the agglomeration effect to foreigninvested companies and enterprises from Hong Kong, China and Taipei,China while the

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 15

FRHIFLHQW IRU VWDWHRZQHG LQFXPEHQWV LV LQVLJQLFDQW LPSO\LQJ WKDW WKH SUHVHQFH RI VWDWH owned enterprises hardly provides investors from Hong Kong, China and Taipei,China ZLWK PXFK LQXHQFH LQ WKHLU ORFDWLRQ FKRLFH In contrast, CPs (7) seem to behave similarly as foreign-invested companies in location decision choice associated with infrastructure-related variables but somewhat differently toward tax incentives. Most domestically invested exporters are normal traders (about   FRQVLGHULQJ FRUSRUDWH LQFRPH WD[ UDWH PRUH VHULRXVO\ WKDQ YDOXHDGGHG WD[ 7KH DJJORPHUDWLRQ HIIHFW DVVRFLDWHG ZLWK IRUHLJQLQYHVWHG UPV LV VLPLODU WR WKDW IRU WKH IRUHLJQRZQHG VXEVDPSOH  DQG WKH HIIHFW RI &3V LV WKH PRVW VLJQLFDQW ZKLOH +7 LQFXPEHQWV KDYH D QHJOLJLEOH HIIHFW +RZHYHU WKH SUHVHQFH RI VWDWHRZQHG UPV KDV D sizable effect on the co-location of CPs. In summary, using results from Models (5) to (7), it can be observed that the presence RI IRUHLJQLQYHVWHG H[SRUWHUV KDV D VLJQLFDQW EHDFRQ HIIHFW ZLWK WKH VWURQJHVW OHDGLQJ UROH LQ DJJORPHUDWLRQ IRU QHZ LQYHVWRUV FRPLQJ IURP WKH 35& +RQJ .RQJ &KLQD 7DLSHL&KLQD DQG IRUHLJQ FRXQWULHV LQ WKH 35&V H[SRUW LQGXVWU\ 0RUHRYHU SULYDWHO\ LQYHVWHG &KLQHVH HQWHUSULVHV VKRZ VRPH WHQGHQF\ IRU FRORFDWLRQ ZLWK VLJQLFDQW DJJORPHUDWLYH LQXHQFH H[HUWHG E\ RWKHU &KLQHVH SULYDWHRZQHG LQFXPEHQW UPV

C.

Processing Traders versus Normal Traders

The last empirical results provide comparison analysis between processing traders and normal traders in different ownership origins. The subsamples in each investment origin are further split into processing traders and normal traders. From Models (8A and 8B) in Table 5, foreign investors in the processing and normal trade sectors behave analogously in location choice with respect to infrastructure and proximity to the coast but the absolute PDJQLWXGH RI HDFK HOHPHQW LV PRUH VLJQLFDQW ZLWK LQYHVWRUV LQ WKH SURFHVVLQJ JRRGV trade sector than those in the normal goods trade sector. As expected, processing traders are more sensitive to an incentive in the form of value added tax reduction than from corporate income tax reduction. Foreign investors in both the processing trade sector and normal trade sector show a tendency to agglomerate with incumbent foreign companies. In particular, the latter likely cluster together with privately-owned Chinese incumbents while the former shows the opposite.

16 | ADB Economics Working Paper Series No. 297

Table 5: Conditional Logit Model: Processing Traders versus Normal Traders (foreign-owned firms)
Model (8): Foreign-Owned A: Processing Traders Variables wage_manufacturing GDP per capita waterway highway high school pop college pop proximity value added tax rate corp income tax rate Incumbent_CP Incumbent_ST Incumbent_HT Incumbent_FN Sample size LL Chi2 Pseudo R2 Estimates Marginal Std. Effects dev. 0.00 0.00 3024 [12.959] [11.644] 0 0 5030 [13.267] [7.353] 5.91 0.33 0.07 [14.255] [12.345] 2.32 0.13 0.21 [7.732] [5.643] 4.81 0.27 0.04 [2.335] [2.745] 8.34 0.46 0.03 [2.966] [2.374] 0.01 0.00 685.8 [21.402] [12.132] 4.66 0.26 0.07 [6.841] [6.874] 3.26 0.18 0.02 [2.544] [2.631] 0.00 0.00 26 [0.126] [0.126] 0.01 0.00 3 [1.152] [1.145] 0.00 0.00 36 [1.373] [1.290] 0.02 0.00 22 [19.780] [8.179] 65309 4903 5436 0.36 Predicted Change* 3.93 2.01 2.34 2.72 1.15 1.55 21.19 1.77 0.41 0.02 0.17 0.15 2.78 B: Normal Traders Estimates Marginal Std. Effects dev. 0.00 0.00 3048 [17.333] [10.857] 0 0 5080 [6.016] [4.571] 8.88 0.28 0.07 [24.640] [10.653] 1.01 0.03 0.21 [4.052] [3.588] 16.07 0.51 0.04 [9.897] [10.730] 21.07 0.67 0.03 [10.824] [13.712] 0.01 0.00 686 [23.885] [10.588] 2.15 0.07 0.07 [3.252] [3.426] 4.16 0.13 0.02 [4.089] [4.005] 0.01 0.00 26 [16.979] [7.414] 0.03 0.00 3 [5.016] [4.362] 0.00 0.00 30 [1.685] [1.550] 0.02 0.00 18 [14.849] [6.570] 78083 6331 5567 0.31 Predicted Change* 2.74 0.51 2 0.68 2.21 2.25 10.01 0.47 0.3 0.73 0.33 0.11 1.04

* Predicted changes (in %) induced by a 1 standard deviation increase in the explanatory variable. Note: Figures in the square bracket are z-statistics. Source: Author's calculations.

Models (9A and 9B) exhibit the results for investors from Hong Kong, China and Taipei,China in processing and normal trade sectors (Table 6). Investors in the processing WUDGH VHFWRU DUH VHQVLWLYH WR YDOXHDGGHG WD[ UDWH ZKLOH WKH FRHIFLHQW RI FRUSRUDWH LQFRPH WD[ UDWH LV LQVLJQLFDQW 6XUSULVLQJO\ LQYHVWRUV IURP +RQJ .RQJ &KLQD DQG 7DLSHL&KLQD LQ WKH QRUPDO WUDGH VHFWRU DUH QRW LQXHQFHG E\ DQ\ WD[ LQFHQWLYH ,QYHVWRUV from Hong Kong, China and Taipei,China in the processing and normal trade sectors seem to agglomerate with incumbent foreign-invested and HT-invested companies. Investors in the normal trade sector show a tendency to choose a province with incumbent Chinese privately invested companies while investors in the processing trade VHFWRU DUH XQOLNHO\ WR RFN WRJHWKHU ZLWK SULYDWHO\RZQHG &KLQHVH LQFXPEHQWV

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 17

Table 6: Conditional Logit Model: Processing Traders versus Normal Traders (Hong Kong, China/Taipei,China firms)
Model (9): Hong Kong, China/Taipei,China-Owned A: Processing Traders Variables wage_manufacturing GDP per capita waterway highway high school pop college pop proximity value added tax rate corp income tax rate Incumbent_CP Incumbent_ST Incumbent_HT Incumbent_FN Sample size LL Chi2 Pseudo R2 Estimates Marginal Std. Predicted Effects dev. Change* 0.00 0.00 3005 16.23 [14.299] [16.707] 0.00 0.00 4905 3.43 [11.095] [10.940] 7.68 0.92 0.07 6.53 [13.934] [13.054] 1.48 0.18 0.21 3.74 [3.163] [3.194] 15.99 1.91 0.04 8.25 [7.984] [9.203] 92.16 11.03 0.03 36.72 [22.706] [39.666] 0.00 0.00 686 29.08 [10.869] [9.425] 3.37 0.40 0.07 2.76 [2.620] [2.652] 5.33 0.64 0.02 1.44 [1.315] [1.286] 0.01 0.00 27 2.19 [5.460] [5.311] 0.01 0.00 4 0.31 [1.036] [1.036] 0.01 0.00 37 3.17 [11.846] [10.067] 0.01 0.00 22 1.37 [3.747] [3.673] 63510 3200 8418 0.57 B: Normal Traders Estimates Marginal Std. Predicted Effects dev. Change* 0.00 0.00 3006 1.8 [3.060] [2.722] 0.00 0.00 4760 1.9 [7.963] [6.771] 8.70 0.86 0.07 6.13 [20.842] [17.879] 0.80 0.08 0.21 1.67 [2.464] [2.427] 1.19 0.12 0.04 0.51 [0.662] [0.679] 19.91 1.97 0.03 6.57 [6.314] [4.947] 0.00 0.00 686 29.84 [18.864] [30.889] 0.55 0.05 0.07 0.37 [0.582] [0.586] 3.10 0.31 0.02 0.7 [1.453] [1.470] 0.01 0.00 28 1.4 [9.438] [8.023] 0.01 0.00 3 0.32 [1.651] [1.639] 0.00 0.00 34 0.87 [4.893] [5.078] 0.01 0.00 19 1.05 [4.109] [3.920] 63140 4698 5346 0.36

* Predicted changes (in %) induced by a 1 standard deviation increase in the explanatory variable. Note: Figures in the square bracket are z-statistics. Source: Author's calculations.

3ULYDWHO\ RZQHG &KLQHVH UPV $ LQ WKH SURFHVVLQJ WUDGH VHFWRU 7DEOH  DUH TXLWH sensitive to corporate tax incentives. Investors in both the processing and normal trade sectors reveal a strong agglomeration effect associated with all types of incumbents except HT incumbents. For processing traders, foreign incumbents have the most VLJQLFDQW DJJORPHUDWLRQ HIIHFW ZKLOH &KLQHVH SULYDWHO\ LQYHVWHG LQFXPEHQWV DJJORPHUDWH normal traders of privately invested entrants (10B).

18 | ADB Economics Working Paper Series No. 297

Table 7: Conditional Logit Model: Processing Traders versus Normal Traders (Chinese private firms)
Variables Model (10): Chinese Private Firms A: Processing Traders B: Normal Traders Estimates Marginal Std. Effects dev. 0.00 0.00 3020 [0.319] [0.323] 0.00 0.00 4928 [6.408] [5.999] 3.13 0.49 0.07 [3.835] [3.343] 1.93 0.30 0.21 [3.629] [3.500] 5.40 0.85 0.04 [1.631] [1.546] 17.92 2.82 0.03 [4.720] [3.975] 0.00 0 686 [9.739] [15.956] 0.79 0.12 0.07 [0.774] [0.782] 5.97 0.94 0.02 [2.686] [2.543] 0.01 0.00 28 [8.399] [7.353] 0.05 0.01 4 [4.570] [4.413] 0.00 0.00 37 [1.297] [1.263] 0.02 0.00 21 [11.235] [8.658] 22489 1808 1636 0.31 Predicted Change* 0.3 4.93 3.51 6.38 3.7 9.47 29.29 0.85 2.12 3.87 3.19 0.64 7.47 Estimates Marginal Std. Effects dev. 0.00 0.00 3008 [11.603] [14.179] 0.00 0.00 4867 [11.353] [8.907] 4.07 0.35 0.07 [16.619] [17.984] 0.46 0.04 0.21 [2.736] [2.659] 9.00 0.76 0.04 [8.144] [10.635] 2.49 0.21 0.03 [1.770] [1.918] 0.00 0 686 [29.888] [20.469] 0.22 0.02 0.07 [0.576] [0.581] 1.39 0.12 0.02 [1.565] [1.543] 0.02 0.00 26 [42.971] [15.156] 0.05 0.00 3 [10.784] [9.132] 0.00 0.00 25 [1.551] [1.597] 0.01 0.00 14 [9.700] [7.667] 186210 15425 12671 0.29 Predicted Change* 3.01 1.95 2.46 0.82 3.32 0.71 14.75 0.13 0.27 4.01 1.3 0.18 1.29

wage_manufacturing GDP per capita waterway highway high school pop college pop proximity value added tax rate corp income tax rate Incumbent_CP Incumbent_ST Incumbent_HT Incumbent_FN Sample size LL Chi2 Pseudo R2

* Predicted changes (in %) induced by a 1 standard deviation increase in the explanatory variable. Note: Figures in the square bracket are z-statistics. Source: Author's calculations.

VI. Policy Implications

For governors in the PRC whose most decisive achievements are judged by increasing gross domestic product in the region, attracting new foreign and domestic direct LQYHVWPHQW LV EHQHFLDO LQ VSXUULQJ HFRQRPLF JURZWK JHQHUDWLQJ PRUH HPSOR\PHQW expanding the tax base, and boosting resident income. Local policy makers encourage WHFKQRORJLFDOO\ VRSKLVWLFDWHG UPV WR HQWHU WKHLU SURYLQFH IRU JUDGXDO LPSURYHPHQW RI production technologies in related industries by means of technology and management knowledge spillover. Correspondingly, policy makers prefer to train and produce a highly

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 19

VNLOOLQWHQVLYH ODERU IRUFH WKDW ZLOO KHOS PHHW WKH GHPDQG IRU KLJKTXDOLW\ MREV DV KLJKO\ sophisticated, technology-based, and R&D-intensive (foreign and domestic) enterprises enter their province. Therefore, provincial planners tend to pay particular attention to WKH HIIHFW RI WKHLU SROLFLHV DQG DFWLRQV RQ WKH EHKDYLRU RI UPV WKDW DUH OLNHO\ WR HQJDJH LQ 5 ' DFWLYLW\ UPV WKDW IRFXV RQ KLJKTXDOLW\ WHFKQRORJ\ GHYHORSPHQW DQG UPV WKDW ZLOO SURGXFH VRSKLVWLFDWHG QDO JRRGV LQ WKH KLJK WHFKQRORJ\EDVHG LQGXVWU\ 7KHUHIRUH they may want to attract foreign-invested enterprises with high technology, as the knowledge and technology transfer induced by their market entry will clearly be welcome externalities. Some explanatory variables analyzed in the empirical section are explicitly and/ RU LPSOLFLWO\ FRQWUROOHG DQG LQXHQFHG E\ SURYLQFLDOOHYHO SROLF\ FKRLFHV ,Q SDUWLFXODU SROLF\ PDNHUV ZRXOG OLNH WR XQGHUVWDQG KRZ YDULDWLRQV LQ WKHLU ORFDO SROLF\ LQXHQFH WKH PDUNHW HQWU\ GHFLVLRQV RI FHUWDLQ JURXSV RI UPV )RU H[DPSOH KLJKZD\V SHU DUHD and length of inland waterways are infrastructure-related policy variables that can be LQXHQFHG E\ EXLOGLQJ PRUH KLJKZD\V PDNLQJ PRUH ULYHUV VXLWDEOH IRU FRPPHUFLDO WUDIF or constructing canals. An upgrade to the waterway and highway infrastructure in the SURYLQFH LV SUHGLFWHG WR KDYH D VLJQLFDQW HIIHFW RQ WKH HQWU\ GHFLVLRQV RI SURFHVVLQJ goods traders (see Model 3 in Table 2). A 1 standard deviation increase in waterways DQG KLJKZD\V ZLOO UDLVH FKRLFH SUREDELOLWLHV E\ DERXW  DQG   UHVSHFWLYHO\ 7KH VKDUH RI KLJK VFKRRO JUDGXDWHV DQG FROOHJH JUDGXDWHV LQ WKH SRSXODWLRQ FRXOG EH LQXHQFHG E\ province-level policies aimed at encouraging higher education. ,W ZRXOG EH IHDVLEOH IRU SURYLQFLDO SODQQHUV WR QG WKH VSHFLF FRVW RI JHQHUDWLQJ D particular infrastructure upgradefor example, building additional highwaysand setting it in relation to the expected effect. For example, suppose that 5,000 new enterprises enter the PRCs export industry every year, and an increase of the choice probability E\  ZRXOG UHVXOW LQ DQ H[SHFWHG DGGLWLRQDO  UPV 7KH H[SHFWHG EHQHWV RI VXFK DGGLWLRQDO LQYHVWPHQW FRXOG EH FRPSDUHG WR WKH FRVW RI VSHFLF SROLF\ LQWHUYHQWLRQV The average wage level in the manufacturing sector is related to policy in important ZD\V WKH IRUPXODWLRQ DQG LPSOHPHQWDWLRQ RI ODERU ODZV LQFOXVLYH RI WKH PLQLPXP ZDJH the restriction of domestic migrant workers (for example, hukou V\VWHP DQG WKH VWDQFH WDNHQ E\ SROLF\ PDNHUV LQ ODERU GLVSXWHV DOO LQXHQFH WKH IDFWRU SULFH RI ODERU 8VLQJ WKH Model (8A), a 1 standard deviation increase to the average wage in manufacturing sector ZLOO UHGXFH FKRLFH SUREDELOLWLHV E\ DERXW   2QH VWDQGDUG GHYLDWLRQ LV  \XDQ SHU year, or 252 yuan per month. This is a sizable wage increase and its predicted effect is relatively large. Overall, investors seem sensitive to the average wage rate in a province. Taking Foxconn for example, the wage level for its low-skilled workers increased by   DQG WKH FRPSDQ\ PRYHG LWV IDFWRU\ IURP 6KHQ]KHQ LQ *XDQJGRQJ SURYLQFH WR Chengdu in Sichuan province in 2011.

20 | ADB Economics Working Paper Series No. 297

5HGXFWLRQ LQ WD[ UDWHV SDUWLFXODUO\ YDOXH DGGHG WD[ EULQJV D VLJQLFDQW LQXHQFH WR investors from foreign countries and Hong Kong, China/Taipei,China in their location choice. An interesting policy implication is given by Models (8) to (10). The agglomeration effect of foreign-invested enterprises seems notably strong for foreign-invested processing traders, HT- invested companies, and private Chinese investors in the processing trade VHFWRU +RVWLQJ IRUHLJQLQYHVWHG HQWHUSULVHV LQ RQH ORFDWLRQ FRXOG DWWUDFW VXEVHTXHQW LQYHVWRUV LQ WKH VDPH ORFDWLRQ ,I WKH SURYLQFLDO JRYHUQPHQW WDUJHWV D VSHFLF LQGXVWU\ WR be developed, then our results suggest that it might kick-start this industry by hosting a number of foreign-invested enterprises to locate in a particular location. The presence of  IRUHLJQLQYHVWHG LQFXPEHQW UPV ZRXOG UDLVH WKH SUHGLFWHG SUREDELOLW\ RI PDUNHW HQWU\ RI RWKHU IRUHLJQ UPV LQ WKH SURFHVVLQJ H[SRUW LQGXVWU\ E\ DERXW   DQG WKDW RI SULYDWH &KLQHVH LQYHVWRUV E\ URXJKO\   7KH 35&V LQGXVWU\ VWUXFWXUH LV VWLOO URRWHG LQ ORZ VNLOOHG ODERULQWHQVLYH PDQXIDFWXUHG JRRGV SURGXFWLRQ DQG KLJKO\ VRSKLVWLFDWHG QDO JRRGV are mostly assembled by foreign-invested enterprises. Policy makers in the PRC need WR QG ZD\V WR DFKLHYH D VXEVWDQWLDO OHYHO RI WHFKQRORJ\ VSLOORYHU IURP IRUHLJQLQYHVWHG enterprises operating in high-technology industry areas to local producers manufacturing LQWHUPHGLDWH JRRGV 7HFKQRORJ\ VSLOORYHU FDQ EH DFTXLUHG WKURXJK LQWHUQDWLRQDOL]DWLRQ RI KRVWLQJ IRUHLJQLQYHVWHG UPV SURGXFLQJ KLJKVNLOOHG WHFKQRORJ\ JRRGV DIWHU ZKLFK ORFDO producers that obtained high-technology skills will be encouraged to gradually replace the foreign-invested processing goods traders through internalization/localization in production of sophisticated manufactured goods. Similarly, the presence of state-owned enterprises in a certain industry will attract more privately invested Chinese companies (Models 10A and 10B) and foreign normal goods traders (Model 8B). These increases only apply to the subpopulation of investors active in WKLV SDUWLFXODU SURGXFW FODVV VR WKDW WKH QXPEHU RI UPV WKDW FDQ EH LQGXFHG WR HQWHU LQ this way may still be small. It suggests that a state-owned infant industry may still play a OLPLWHG OHDGLQJ UROH WR GHYHORS VSHFLF WDUJHW LQGXVWULHV

VII. Conclusion
The most important result of this study is the proof of a basic concept: that heterogeneity LQ UP RZQHUVKLS DQG WUDGH W\SH FDQ HQKDQFH WKH XQGHUVWDQGLQJ RI LQYHVWPHQW ORFDWLRQ decisions in the PRC. Processing traders differ substantially from nonprocessing traders in response to changes in wage cost, level of development in infrastructure, and proximity WR WKH 3DFLF FRDVW 3UHVHQFH RI IRUHLJQLQYHVWHG HQWHUSULVHV FUHDWHV D EHDFRQ HIIHFW RQ VXEVHTXHQW IRUHLJQ DQG GRPHVWLF LQYHVWRUV 7KH PRVW EDVLF FRQFOXVLRQ IURP WKH HPSLULFDO DQDO\VHV LV WKDW SUR[LPLW\ WR WKH 3DFLF coast and new economic geography of agglomeration effect and the presence of LQFXPEHQW UPV LQ WKH VDPH SURGXFW FODVV DUH WKH VWURQJHVW GHWHUPLQDQWV RI DQ LQYHVWRUV

New Economic Geography and Tax Competition in the PRC: A Firm-Level Data Analysis with Policy Implications | 21

location choice in the PRCs export industry. Agglomeration also plays a greater role for investors specializing in processing goods trade and highly skill-intensive or R&D intensive production. Firms with a higher educated workforce agglomerate more than others. 6LPLODUO\ SURFHVVLQJ WUDGH UPV GLVSOD\ VWURQJHU DJJORPHUDWLYH EHKDYLRU WKDQ nonprocessing traders. Moreover, processing traders are more sensitive to the wage level and value proximity to the coast more. A wide range of industrial policies designed WR DWWUDFW FOXVWHUV RI OHDGLQJ UPV PD\ EH MXVWLDEOH RQ WKHVH JURXQGV $PRQJ LQYHVWRUV of different origin, Hong Kong, China and Taipei,China investors seem the least driven by agglomeration. Investors from Hong Kong, China and Taipei,China, by contrast, tend WR FRORFDWH ZLWK RWKHU UPV IURP WKH VDPH UHJLRQ ZKLOH IRUHLJQLQYHVWHG UPV VHHP WR EH QHJDWLYHO\ LQXHQFHG E\ WKH SUHVHQFH RI +RQJ .RQJ &KLQD7DLSHL&KLQD LQFXPEHQWV )RUHLJQLQYHVWHG LQFXPEHQW UPV KDYH D YHU\ ODUJH HIIHFW RQ VXEVHTXHQW SULYDWH &KLQHVH investors in processing trade sector, and have a sizable effect on other foreign investors and investors from Hong Kong, China and Taipei,China. %HQHWV IURP UHGXFWLRQ LQ YDOXHDGGHG WD[ UDWHV DQG FRUSRUDWH LQFRPH WD[ UDWHV SOD\ an evident role for foreign investors and HT investors in the processing trade sector. However, since 2008, a number of provinces started dismantling preferential labor laws, export subsidies, favorable value-added tax rates, and corporate tax rates to lowskill-intensive foreign enterprises including processing goods traders. Policy makers have publicly stated their goal that the PRC will dynamically transform its comparative advantage from low-skill-intensive to high-tech industry concentrated. Policy makers in the PRC intend to transform the internationalization process through attraction of foreign investment toward localization by reducing the processing good trading portion and by increasing the supply of domestically produced intermediate inputs for their exports. As the PRC moves up to the high value supply chain, expands its domestic market, and exports skill-intensive goods, processing goods traders might consider new places or new business models.

References
Amiti, M. and B. S. Javorcik. 2008. Trade Costs and Location of Foreign Firms in [the Peoples Republic of] China. Journal of Development Economics 85:12949. Amity, M., and C. Freund. 2010. An Anatomy of [the People's Republic of] Chinas Export Growth. In R. Feenstra and S-J Wei, eds., [The Peoples Republic of] Chinas Growing Role in World Trade. National Bureau of Economic Research, Massachusetts. Belderbos, R., and M. Carree. 2002. The Location of Japanese Investment in [the Peoples Republic of] China: Agglomeration Effects, Keiretsu, and Firm Heterogeneity. Journal of Japanese and International Economies 16:194211.

22 | ADB Economics Working Paper Series No. 297

Boermans, M., H. Roelfsema, and Y. Zhang. 2009. Regional Determinants of FDI in [the People's Republic of] China: A New Approach with Recent Data. Tjalling C. Koopmans Research Institute Discussion Paper Series No. 09-23, Utrecht School of Economics, The Netherlands. Brueckner, J. K. 2002. Strategic Interaction among Governments: An Overview of Empirical Studies. International Regional Science Review 44:43867 Brueckner, J. K., and L. Saavedra. 2001. Do Local Governments Engage in Strategic Property Tax Competition? National Tax Journal 54(2):20329. Cameron, A. C., and P. K. Trivedi. 2005. Micro-econometrics: Methods and Applications. Cambridge: Cambridge University Press. Carlton, D. W. 1983. The Location and Employment Choices of New Firms: An Econometric Model with Discrete and Continuous Endogenous Variables. Review of Economics and Statistics 65:44049. Chang, S-J, and S. Park. 2005. Types of Firms Generating Network Externalities and MNCs Colocation Decisions. Strategic Management Journal 26(7):595615. Chen, G. S. 2009. Agglomeration Economies and the Location of [Taipei,China] Investment in [the Peoples Republic of] China. MPRA Paper No. 13896, Monash University, Munich. Cheng, L. K., and Y. K. Kwan. 2000. What are the Determinants of the Location of Foreign Direct Investment? The Chinese Experience. Journal of International Economics 51(2):379400. Head, K., and J. Ries. 1996. Inter-City Competition for Foreign Investment: Static and Dynamic Effects of [the Peoples Republic of] Chinas Incentive Areas. Journal of Urban Economics 40:3860. Hu, A. G., and R. F. Owen. 2005. Gravitation at Home and Abroad: Regional Distribution of FDI in [the Peoples Republic of] China. Department of Economics, National University of Singapore. Mimeo. Krugman, P. 1991. Increasing Returns and Economic Geography. Journal of Political Economy 99:48399. McFadden, D. L. 1974. Conditional Logit Analysis of Qualitative Choice Behavior. In A. =DUHPEND / /XQGTYLVW ) 6QLFNDUV DQG - :HLEXOO HGV Frontiers in Econometrics. New York: Academic Press.   0RGHOLQJ WKH &KRLFH RI 5HVLGHQWLDO /RFDWLRQ ,Q $ .DUOTYLVW / /XQGTYLVW ) Snickars, and J. Weibull, eds., Spatial Interaction Theory and Planning Models. North Holland: Amsterdam. Qian, Y., and G. Roland. 1998. Federalism and the Soft Budget Constraint. American Economic Review 88(5):114362. Rodrik, D. 2006. Whats so Special about [the Peoples Republic of] Chinas Exports? NBER Working Paper No. 11947, National Bureau of Economic Research, Cambridge. Wakasugi, R. 2005. The Effects of Chinese Regional Conditions on the Location Choice of -DSDQHVH $IOLDWHV Japanese Economic Review 56(4):390407. Wang, Z., and S. Wei. 2008. What Accounts for the Rising Sophistication for [the People's Republic of] Chinas Exports? Working Paper 13771, National Bureau of Economic Research, Cambridge. Wilson, J. D. 1999. Theories of Tax Competition. National Tax Journal 52(2):269304. Woodward, D. P. 2006. Beyond the Silicon Valley: University R&D and High-technology Location. Journal of Urban Economics, 60:1532.

About the Paper Minsoo Lee analyzes the spatial determinants of new foreign and domestic enterprises in the export industry of the Peoples Republic of China (PRC) by applying new economic geography of agglomeration theory, tax competition theory by resource flow model, and comparative advantage analysis. He finds that agglomeration plays a greater role for foreign investors specializing in the processing goods trade and highly skillintensive or research and development-intensive production, and that benefits from reduction in the value added tax rate and corporate income tax rate play an evident role for foreign investors and Hong Kong, China and Taipei,China investors in the processing trade sector. As the PRC moves up the high-value supply chain and transforms the internationalization process through attraction of foreign investment toward localization by reducing the processing goods trading portion and increasing the supply of domestically produced intermediate inputs for their exports, processing goods traders may consider new locations or business models.

About the Asian Development Bank ADBs vision is an Asia and Pacific region free of poverty. Its mission is to help its developing member countries reduce poverty and improve the quality of life of their people. Despite the regions many successes, it remains home to two-thirds of the worlds poor: 1.8 billion people who live on less than $2 a day, with 903 million struggling on less than $1.25 a day. ADB is committed to reducing poverty through inclusive economic growth, environmentally sustainable growth, and regional integration. Based in Manila, ADB is owned by 67 members, including 48 from the region. Its main instruments for helping its developing member countries are policy dialogue, loans, equity investments, guarantees, grants, and technical assistance.

Asian Development Bank 6 ADB Avenue, Mandaluyong City 1550 Metro Manila, Philippines www.adb.org/economics ISSN: 1655-5252 Publication Stock No. WPS124550
January 2012

< 0124 5509 >

Printed on recycled paper Printed in the Philippines