Value-Added Taxes and International Trade

Value-AddedTaxesandInternationalTrade:TheEvidenceMihirA.DesaiHarvardUniversityandNBERJamesR.HinesJr.UniversityofMichiganandNBERNovember2002FinancialsupportfromtheInternationalTaxPolicyForumisgratefullyacknowledged.Value-AddedTaxesandInternationalTrade:TheEvidenceABSTRACTThispaperexaminestheeffectofvalue-addedtaxes(VATs)oninternationaltrade.Destination-basedVATsarecommonlythoughttoencourageexports,sinceexportsareexemptfromtaxwhileimportsaretaxed.Economictheoryimpliesthatexchangerateadjustmentpreventsdestination-basedVATsfromaffectingexportsandimports,sinceexchangerateappreciationcompletelyundoestheeffectsofintroducingaVAT.Indeed,thispropositionissowellacceptedamongeconomiststhatithasnotbeensubjectedtoseriouspriortesting.Evidencefrom136countriesin2000indicatesinsteadthatrelianceonVATsisassociatedwithfewerexportsandimports.CountriesusingVATshaveone-thirdfewerexportsthandocountriesnotusingVATs,and10percentgreaterVATrevenueisassociatedwithtwopercentfewerexports.Asimilarpatternappearsinanunbalancedpanelof168countriesfrom1950-2000,inwhichVATuseisassociatedwith12percentfewerexports.Thesepatternspersistwiththeinclusionofincomeandgeographiccontrols,andwhiletheeffectofVATsonexportsisstrongeramonglow-incomecountriesthanitisamonghigh-incomecountries,thereisasignificantnegativeeffectofVATsonexportsevenamonghigh-incomecountries.ThebehaviorofAmericanmultinationalfirmsin1999isconsistent:tenpercentgreaterlocalVATcollectionsareassociatedwith5percentfewerexportsbylocalAmerican-ownedaffiliates.TwofeaturesofVATimplementationmayaccountfortheseeffects:VATstendtobeimposedathigherratesontradedgoodsthanonnontradedgoods,andexportersoftenreceiveonlyincompleteVATrebates.JELClassifications:H87,H25,F17,F23.MihirA.DesaiJamesR.HinesJr.HarvardBusinessSchoolUniversityofMichiganMorgan363BusinessSchoolSoldiersField701TappanStreetBoston,MA02163AnnArbor,[e-mail]@umich.edu11.IntroductionValue-addedtaxes(hereinafter,VATs)representimportantsourcesofgovernmentrevenueinmostoftheworldoutsidetheUnitedStates.Overthelasthalf-century,VATshavebeenadoptedinthevastmajorityofdevelopedanddevelopingeconomiesand,fortheseeconomies,provideanaverageofonequarterofgovernmentrevenue.WhileVATsdifferindetail,theircommonstructureisthattheytaxfinalconsumptionbyimposingtaxesonactivities(“valueadded”)alongtheproductionchain.OnecanreasonablythinkofaVATasasensibleformofasalestax,withtheimportantdistinctionthataVATdoesnottaxsalesofintermediategoodsfromonebusinessunittoanother.Sincetheseintermediategoodsincludecapitalinvestmentgoods,VATsdonottax,andtherebydiscourage,capitalinvestmentsinthewaythatcorporateincometaxesdo.Asaresult,VATsarethoughttobeveryefficientdevicesforraisinggovernmentrevenue,whichhascontributedtotheirgrowingpopularity.Mostoftheworld’sVATsaredestination-based,whichmeansthatthetaxesareimposedonlyongoodsandservicesconsumedwithinthetaxingjurisdiction.Underaproperly-workingdestination-basedVAT,exportsreceiverebatesequaltotheamountofVATpaidinthecourseofproducingtheexporteditem,whileimportsaresubjecttotheVATatthesamerateasdomestically-producedgoods.Sinceadestination-basedVATisataximposedonimportsandonotherdomestically-consumedgoods,andeffectivelynotimposedonexports,itisacommonplacebeliefthataVATencouragesexports,possiblyatthecostofimportsanddomesticconsumption.Thetheoryofinternationaltradeoffersaverydifferentprediction.Intheory,thedestination-basednatureofaVATshouldhavenoeffectwhatsoeveronexportsandimports.2ThereasonisthatexchangeratesadjusttoundotheeffectsofVATsonincentivestoexportandimport.Ifacountryimposesan8percentdestination-basedVATonallgoodsandservices,anditsexchangerateappreciatesby8percentasaresult,thenreturnstoimportersandexporterswillnotchange.Thiskindofone-for-oneadjustmentofexchangeratestoadestination-basedVATisexactlywhattheorypredicts,andasaconsequence,VATsarenotexpectedtoinfluencepatternsofinternationaltrade.Inpractice,VATsmaysubstituteforothertaxes(suchascorporateincometaxes)thataffectimportsandexports,VATsareoftennonuniformlyappliedtogoodsandservices,andtherearesomedifficultiesinVATadministrationthatpreventexportersfromreceivingVATrebates–allofwhichhavethepotentialtoinfluenceexportbehavior.Astheseconsiderationsaretypicallythoughttobeofminorimportance,thereceivedwisdomofinternationaltradetheoryisthatdestination-basedVATsshouldaffectexchangeratesbutnotimportsorexports.Thistheoryissowidelyacceptedthatithasseldom,ifever,beensubjectedtotestsbasedonactualexperience.ThepurposeofthispaperistoconsiderevidenceoftheeffectofVATsoninternationaltradepatterns.TheresultsareconsistentneitherwiththecommonplaceviewthatVATsencourageexports,norwiththetheoreticalpredictionthatVATsdonotaffecttrade.Instead,theevidenceindicatesthatcountriesrelyingonVATshavefewerexportsandfewerimports(asafractionofGDP)thandocountriesrelyingontaxesotherthanVATs.CountriesusingVATshaveone-thirdfewerexportsthanthosenotusingVATs,aftercontrollingforincomeandvariousgeographicattributes.ThesurprisingnegativecorrelationbetweenVATrelianceandinternationaltradeisstrongestamonglow