交通运输基础设施外文文献.docx

交通运输基础设施外文文献.docx

《交通运输基础设施外文文献.docx》由会员分享，可在线阅读，更多相关《交通运输基础设施外文文献.docx（13页珍藏版）》请在冰豆网上搜索。

交通运输基础设施外文文献

文献信息

标题:

BilateralTransportCost,Infrastructure,CommonBilateralTiesandPoliticalStability

作者:

Molina,Danielken

期刊:

RevistadeEconomíadelRosario

卷:

11；期:

2；页:

221-247；年份:

2015

BilateralTransportCost,Infrastructure,CommonBilateralTiesandPoliticalStability

Molina,Danielken

Abstract

TransportCostsareoneofthemostimportantsourcesofbarrierstotrade.Inspiredbythisconcern,weuseanad-hocempiricalmodeltoexaminethepossibledeterminantsoftransportcosts.

Usingacountryaggregatebilateraltransportcostsdatabasefor1990,wereplicateLimaoandVenables（2001）andwefindthattheeffectoftheirinfrastructureindexontransportcostsisnotrobust.FollowingMicco（2004）andMiccoandSerebrinzky（2005）wecalculatetwoalternativeindexesofinfrastructure.Usingthesenewindexeswefindthatnotonlydistancebutinfrastructure,politicalstability,commonbilateraltiesandopenskyagreementsaswellareotherimportantchannelsthroughwhichtransportcostscanbereduced.

Keywords:

transportcosts,infrastructure.

1.Introduction

TransportCostsareoneofthemostimportantsourcesofbarrierstotrade.AsObstfeldandRogoff（2001）explain,severalpuzzlescanbeexplainedbytheirexistence.ButasHummels（2006）pointsout,theliteraturehasnotyetdevotedenoughattentiontotrytomodelthedeterminantsoftransportcosts.Previousapproacheshavemodeledthem,buttheapproachesarecompletelyadhocandtheyarecharacterizedbyheuristicfunctionalspecifications.Therefore,webelievethatfutureworkcanbedevelopedinthisarea.

Giventhisconcern,weuseanad-hocempiricalmodeltoexaminethepossibledeterminantsoftransportcosts,inspiredspeciallybythefactthattheyhavebeendecreasingsincetheearly1980's.1Inparticular,weestimatetheeffectofinfrastructureanddistanceonbilateraltransportcosts.Usingacountryaggregatebilateraltransportcostsdatabasefor1990,LimaoandVenables（2001）（henceforthLV）estimatethatthedeteriorationoftheinfrastructureindexofacountryfromthemedianleveltothe75thpercentileraisestransportcostsby12percentagepoints.Thiseffectimpliesareductionintradevolumebyaround28percent.2

Followingthesameapproach,wereplicateLVresults,andwefindthattheeffectofhisinfrastructureindexontransportcostsisnotrobust.AdoptingMicco（2004）andMiccoandSerebrinzky（2005）wecalculatetwoalternativemeasuresofinfrastructurewhichweconsiderareabetterproxyoftheportinfrastructureofacountry.Inaddition,weestimatetheeffectofcommonbilateralties,countrypoliticalstabilityandopenskyagreementsontransportcosts.

Wefoundthatinadditiontodistance,infrastructure,politicalstability,commonbilateraltiesandopenskyagreementsareotherimportantchannelsthroughwhichtransportcostscanbereduced.

Thestructureofthispaperisasfollows:

SectionIIsetsupthemodelandtheempiricalestimationstrategythatLVused.SectionIIIperformsadetaileddescriptionoftherequireddata.Inparticular,thesectiondescribeshowwecalculatetheinfrastructuremeasures.SectionIVdescribetheresultsobtained.First,wereplicatetheresultsobtainedbyLVintables2,3,5,6,7and8.Secondweexpandthedata,andweperformpaneldataestimatesoftheeffectswiththeLVinfrastructureindex,andournewinfrastructurevariablesandthenewsetsofcontrols.SectionVconcludes.

2.TheModel

FollowingLV,weknowthattheunitcostsofshippingagood（Tij）fromcountryitocountryjinperiodtisgivenbythefollowingequation:

3

Wherexijisavectorofcharacteristicsrelatedtothejourneybetweencountryiandcountryj.Xiisavectorofcharacteristicsrelatedtocountryi.Xjisavectorofcharacteristicsrelatedtocountryj,andìijisavectorofnonobservables.

Followingthetradeliterature,4LVusedthebilateraldistancebetweencountryiandcountyjandaborderdummyvariableasthejourneyspecificvariablesdefinedinxij.Asexpected,thehigherthedistancebetweenthetwodestinationsthehigherthetransportcostare,andsecond,whentwocountriesshareacommonborder,wewouldexpectthatotherfactorsbesidesdistancemightreducetransportcosts.

First,neighboringcountriesmighthavemoreintegratedtransportnetworksthatreducetransportcosts.Second,neighboringcountriesmighthavesimilarcustomsagreementsortransitrulesthatdecreasetheamountoftimeexpendedintransportinggoodsbetweendestinations.Thisimpliesareductionintheshippingandinsurancecostschargedpergood.Third,thehigherthetradevolumeamongneighboringcountries,thelowerthefixedcostsare.5Thereasonisduetothefactthatthehighvolumeoftradebetweentwolocationsreducesthefixedcostssharedbythesenderswithinthetwodestinationpointsbecauseofcargobackhauling.6

AsforthecountryspecificcharacteristicsdefinedinXiorXj,LVonlyfocustheirattentionongeographicalandinfrastructuremeasures.Forgeographicalcharacteristics,LVcontrolsifeitherthecountryisanislandornot,andtheyalsocontrolwhetherthecountryislandlockedornot.7

Themotivationtofocusonthesetwomeasuresisgivenbythefactthatlandlockedcountriesmighthavehighertransportcostssincetheydonothaveports;therefore,cargomustfirstpassthroughneighboringcountriesfirst,whichincreasesthetimeoftransportationandinconsequenceincreasestransportcosts.Inrelationtoislands,theyexpectthatislandsmayhavebetterportsinfrastructurewhichimpliesareductionintransportcosts.

SinceLVuseinformationatcountrylevel8obtainedfromtheDirectionalTradeData（DOTS）,theadvaloremtransportcostfactorbetweencountryiandcountryjisdefinedastheratiobetweenC.I.FandF.O.Btradevalues.9Then,ourmeasureoftransportcostsisdeterminebythefollowingequation:

Undertheassumptionthattijcanbeapproximatedbyaloglinearfunction,wecanproxytransportcostfactorsbetweencountryiandcountryjbythefollowingequation:

10

Whereweassumethat!

jisuncorrelatedwiththeexplanatoryvariablespreviouslydefinedinxij,XiandXj.Aswediscussinthefollowingsection,thereareacoupleoffactorsthatshouldbetakenintoaccountbeforeadvancingtoestimateequation（3）.First,theresultsobtainedbyLVarebasedonthestrongassumptionthattransportcostsarealinearfunctionofthevariablesdefined.Thiscanbetestedbyintroducingsomenonlinearrelationshipswiththeinfrastructurevariablepercountry.SincewearefollowingLV,wedonottestforthenonlinearrelationshipoftheinfrastructureindexes.Second,Yeats（1978）andHummels（2006）pointouttransportcostdataproxybythematchedpartnertechnique11hasanimportantportionofitsownvarianceexplainedbynoise.Third,sinceDOTSdataisobtainedfromthreedifferentsources,itissubjecttodatadifferencesthatarenotrelatedtotradecosts.Certainstatisticalofficesmayvaluegoodsdifferently,i.e.fluctuationoftheexchangeratewithinthetimethatittakestotransportthegoodbetweenlocationmaydeterminedifferentvaluationmethodsforproductsinC.I.FandFOBvaluesrespectively.Fourth,discrepancyinlevelsofC.I.FandFOBamongmatchedcountrydatacaninducetohighadvaloremtransportcosts.Fifth,DOTSdataforasingleyearforagivencountrymightchangewithindifferentpublications.Althoughthechangeissmallinlevels,it'simplicationontheadvaloremtransportcostrate（asdefinedinequation2）couldbeimportant.

3.Data

Sincethispaperreplicatestables2,3,5,6,7and8fromLV,thevariablesthatweneedtoreplicatetheresultsobtainedbytheauthorsare:

advaloremtransportcosts,bilateraldistancebetweencountriesiandj,commonborderdummy,islanddummyforcountriesiandj,GDPandGDPpercapitaforcountriesiandj,infrastructureindexforcountriesiandj,landlockeddummy,Africancountrydummyforcountriesiandjandruleoflawandcontrolofcorruptionofcountriesiandj.

Advaloremtransportcosts（tij）definedbyequation

（2）weredirectlycalculatedfromDOTStradedataforyear2005.AsexplainedbyHummels（2006）,thisdatahastheadvantagethatithasaverygoodcoverage,ithasinformationofbilateralimportsandexportsatC.I.FandF.O.Btradevaluesforalmostallthecountriesoftheworldfrom1948uptothepresent（withalagoftwoyears）.Butaswementionedbefore,ithassomeproblemsaswell.TheIMFbuildsupthisdatafromseveralsources,whichimpliesthatthedatahasvariationsthatarenotduetotransportcosts.Sowhenweproxytijbyequation

（2）,partofthevariationintijisnotduetotransportcostsperse.

Differencesacrosscountriescanbeduetodifferentvaluationsoftrade;i.e.atF.O.Bacountrycanreportvalueofgoodintheship,andanothercanreportthevalueofthegoodbeforeenteringtheship,theU.S.valuesimportsattheexchangerateduethesamedaytheproductentersthefirstport,butothercountriesusetheaverageexchangeratewithintheperiodthatittookthegoodtoarrivefromcountryitocountryj.Otherdifferencescanbeexplainedbythecountryqualityofthedata.Thisbiasisparticularlyimportantfordevelopingcountrieswherethequalityofdataisusuallylowerthanindevelopedcountries.

Anothersourceofdifferencesarethere-adjustmentsmadebytheIMFtoclarifyerrorsandmistakesfrompreviousyears.Soeventhoughweaccesstherawdatatocalculatetij,ourmeasurebydefinitionisdifferentfromthemeasureusedbyLVbecauseIMFhascorrectedthenumbers.

Inaddition,IMFhasaruleofimputingvaluesforimportsandexports.Ifdataisavai