Dispersion forms.an interaction of market access,

JOURNALOFREGIONALSCIENCE,VOL.49,NO.1,2009,pp.177–204DISPERSIONFORMS:ANINTERACTIONOFMARKETACCESS,COMPETITION,ANDURBANCOSTS*HajimeTakatsukaGraduateSchoolofManagement,KagawaUniversity,Saiwai-cho2-1,Takamatsu,Kagawa760-8523,Japan.E-mail:[e-mail]-ZhiZengGraduateSchoolofInformationSciences,TohokuUniversity,Aoba6-3-09,Aramaki,Aoba-ku,Sendai,Miyagi980-8579,Japan;andCenterforResearchofPrivateEconomy,ZhejiangUniversity,Zhejiang310027,China.E-mail:[e-mail]-regionmodelincludingmultipleindustrieswithdifferenttransportcosts.Tworesultsarederived.First,dispersionoccursforsmalltransportcosts,butthespecificdispersionpatternsdependonthelevelofurbancosts.Thisresultsfromaninteractionofthemarket-accesseffectonconsumers,themarket-accesseffectonfirms,thecompetitioneffect,andtheurban-costeffect.Second,decreasingtransportcosttendstoletindustrieswithlowertransportcostsdisperse,althoughthesharesofindustrieslocatinginthelargerregionarenotinorderoftheirtransportcosts.Wefurtherprovidesomeempiricaldataconcerningthesecondresult.1.INTRODUCTIONDuetothe20thcenturytechnologicalinnovations,thetransportcostsformanufacturedgoodshavedeclinedmorethan90percent(GlaeserandKohlhase,2004).Theliteratureofneweconomicgeography(NEG)showsthatindustriallocationisstronglyrelatedtodecreasingtransportcosts(Fujita,Krugman,andVenables,1999;FujitaandThisse,2002).Specifically,thetwo-regionmodelofKrugman(1991)showsthatthemanufacturingindustrydis-persesequallyintwosymmetricregionswhenthetransportcostsarehighandthemanufacturingindustryagglomeratesinoneregionwhenthetrans-portcostsaresmall.IncludingurbancostsinKrugman’score-peripherymodel,Helpman(1998)andTabuchi(1998)showthatthedispersingequilibriumbe-comesstableagainifthetransportcostsareverysmall.*TheauthorsareverygratefultoYoshitsuguKanemoto,TomoyaMori,HiroshiOhashi,RyosukeOkamoto,DanSasaki,KomeiSasaki,TakatoshiTabuchi,andanonymousrefereesaswellasMarkPartridgefortheirusefulcomments.ThefinancialsupportfromtheJapanSocietyforthePromotionofSciencethroughGrants-in-AidforScienceResearch20730183forthefirstauthorand18530179and19203013forthesecondauthorareacknowledged.ThesecondauthoralsoacknowledgestheCRPEjointresearchsupportfromZhejiangUniversity.Received:June2007;revised:November2007;accepted:April2008.C2008,WileyPeriodicals,Inc.DOI:10.1111/j.1467-9787.2008.00588.x177178JOURNALOFREGIONALSCIENCE,VOL.49,NO.1,2009Theabovechangeoflocationpatternsinthecore-peripherymodelcanbeexplainedastheresultoftheinteractionoffoureffects,namely,themarket-accesseffectonconsumers,themarket-accesseffectonfirms,thecompetitioneffect,andtheurban-costeffect.Thetwomarket-accesseffectsdescribethetendencyofconsumersandfirmstolocateinalargerregioninordertosavethetradecosts;thus,theyencourageagglomerationandincreasewithtransportcosts.Thecompetitioneffectreferstothetendencyoffirmstolocateintheregionwithfewerfirmsinordertoavoidseverecompetition;therefore,italsoincreaseswithtransportcostsbutworksagainstagglomeration.Urbancostsincludecommutingandhousingcosts,whichrisewiththenumberofresidents,sotheurban-costeffectdiscouragesagglomeration.Toseehowthefoureffectsworkanddeterminethelocationpattern,con-siderthesituationthateachfirmrequiresacertainnumberofmobile(skilled)workers,andthattherearenourbancosts.Ifallworkersaremobile,agglom-eratinginoneregionmustbestable.Thisisbecauseevenifsomeofthemleavetheagglomeratedregiontotheotherregion,theycouldnotfindcustomersexceptforthemselves.Inotherwords,themarket-accesseffectsdominatethecompetitioneffect.However,Krugman(1991)showsthatthecompetitioneffectcouldsurpassthemarket-accesseffectsand,thus,firmsandtheirworkersdis-perseiftherearemanyimmobileworkersinbothregionsandtransportcostsarehigh.Ontheotherhand,Helpman(1998)andTabuchi(1998)showthat,ifthereareurbancostsandthetransportcostsarelow,theurban-costeffectwouldberelativelylargesinceitdoesnotdecreasewhenthetransportcostsfall,and,thus,firmsandtheirworkerswoulddisperseagain.Suchaprocessiscalledredispersiontobedistinguishedfromthedisper-sionduetolargetransportcostsintheliterature.Theredispersionprocesscanalsobederivedfromothersettings.First,thepositivetransportcostsfortraditionalgoodscouldcauseredispersion(Fujitaetal.,1999,Chap.7;PicardandZeng,2005).Thereasonisthatatlowleveloftradecostsofmanufacturinggoods,workerswanttobewheretraditionalgoodsarecheaper,sotheyspreadacrossregionsagain.Second,immobilityofworkerswithinput-outputlinkagesbetweenfirmscouldcauseredispersionoffirms(Venables,1996;Puga,1999).Withhighbarrierstomigrationlikeaninternationalcontext,atlowleveloftradecostsofmanufacturinggoods,firmswanttobewhereimmobilefactorsarecheaper,sotheyspreadacrossregionsagain.Theseminalpaperscitedaboveassumeasinglemanufacturingindus-try.Locationpatternsandtheinteractionoffoureffectsbecomemuchmorecomplicatediftherearemultipleindustries.Especially,itisnoteworthythatthemarket-accesseffectonfirmsandthecompetitioneffectdependonthetransportcostsandthatindustriesfacedifferentlevelsoftheseeffectsiftheyhavedifferenttransportcosts.