IMF-國際稅收溢出與有形投資-對全球最低稅收的啟示（英）-2023.8

Investment,WithImplicationsfortheaelKeenLiLiuandHayleyPallanWP9IMFWorkingPaperspasoJiqaJasaeJouindJo6Jassq(\uaeu\uoJ)s(enpeJaduqIisuap\oaIioi\oomman\senp\oanoouJe6apaqe\a.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.2023AUG*Keen:TokyoCollege,UniversityofTokyo;InstituteforFiscalStudies,London;CERDI,UniversiteClermontAuvergne;CenterforBusinessTaxation,UniversityofOxford;Liu:InternationalMonetaryFund(lliu@).Pallan:WorldBank(hpallan@).WearegratefultoRuuddeMooij,MichaelDevereux,DinarPrihardini,AndrzejStasioandparticipantsinseminarsattheIMF,Mannheim,NBER,andOxfordforhelpfulcomments.Allmistakesareourown.Theviewsexpressedaretheauthors'anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.Thefindings,interpretationsandconclusionsexpressedinthispaperareentirelythoseoftheauthorsandshouldnotbeattributedtotheWorldBank,itsExecutiveDirectors,orthecountriestheyrepresent.?2023InternationalMonetaryFundWP/23/159IMFWorkingPaperFiscalAffairsDepartmentInternationalTaxSpilloversandTangibleInvestment,WithImplicationsfortheGlobalMinimumTaxPreparedbyMichaelKeen,LiLiu,andHayleyPallan*AuthorizedfordistributionbyAlexanderKlemmAugust2023IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.ABSTRACT:Thispaperarticulatesand,usingnewly-assembleddata,exploreshowinternationaltaxationaffectsaggregatetangiblecross-borderinvestment.Spilloversfromstatutorytaxratesabroadseem:Assizableaseffectsfromthehost’srate;largerthanpreviousconsensusvalues(attributedtoasystematicbiasfromFDIdata);andconsistentwith‘implicit’profitshiftingthroughrealinvestment(ratherthan‘paper’profitshifting).Contrarytomuchpolicydiscussion,theresultsalsoimplythat:Hostcountries’marginaleffectivetaxrateshaveatbestaweakeffectonrealinvestment;thoseelsewherehavenone;and,appliedtotheprospectiveglobalminimumtax,inwardtangibleinvestmentinmostsamplecountrieswillincrease.JELClassificationNumbers:H21,H25,H32Keywords:CorporateTaxation;InternationalTax;MultinationalInvest-ment;ForeignDirectInvestmentAuthor’sE-MailAddress:michael.keen@mail.u-tokyo.ac.jp;lliu@;hpallan@3Contents1Introduction2ChannelsofTaxSpillover3ForeignAffiliateInvestment3.1UnderstandingForeignAffiliateInvestment(andFDI)data......4EmpiricalStrategySpecificationandOtherData...............5Cross-borderTaxEffectsonInwardFAIults5.5ComparingTaxEffectsinFAIandFDIData..............6TheGlobalMinimumTaxandTangibleInvestment7ConclusionsAProofofProposition1BSufficientconditionforaminimumtaxtoincreaseaggregateFAICSupplementaryTablesandFigures715404669014ListofFigures1FDIandFAI:Anexample 482FAIvs.FDITotalHostsinSamplevs.EUHostSample 493FAIvs.FDI:ByParentEconomyCharacteristics 504Non-linearEffectsofCorporateTaxRates 515DistributionofInvestmentImpactandMinimumTaxation 526Numbersof‘Winners’and‘Losers’atVariousMinimumRates 537AggregateFAIandMinimumTaxRates 54C.1Worldwidevs.TerritorialParentCITSystemsandtheRateDifferential71C.2Worldwidevs.TerritorialParentCITSystemsandtheDifferentialTermComponents 72SizeandCorporateTaxRatesin C.4HeterogeneitybyCorporateIncomeTaxVariables 745ListofTables1FDIvs.FAI-StylizedComparison 191SummaryStatistics 552BaselineResults 563SpilloversfromKF-METRsoutsidetheHostCountry? 574‘Pure’ProfitShifting 585TaxationintheParentCountry:Worldwidevs.Territorial 586HeckmanSelectionModel 597MoreComplexDynamicofInvestmentandTimingofTaxResponse.608AlternativeWeightingSchemesinConstructingtheRateDifferentialTerm 619Endogeneity:AControlFunctionApproach 6210Country-PairFixedEffects 6311Cross-borderTaxEffects:FAIvs.FDI 64C.1SummaryofMainTaxVariablesbyCountry 75C.2NumberofParentCountriesbyHostCountryandYearsintheData76C.3AllParentsintheDataset:WorldwideandTerritorialClassification.77CTaxRatesin2016 78C.5DistributionoftheStatutoryRateDifferentialTermin2016 796BEABureauofEconomicAnalysisBaseErosionandProfitShiftingCITCorporateIncomeTaxEUEuropeanUnionFAIForeignAffiliateInvestmentForeignAffiliateStatisticsFDIForeignDirectInvestmentGroupof20GloBEGlobalAnti-BaseErosionMETRMarginalEffectiveTaxRateMNCMultinationalCorporationOECDOrganisationforEconomicDevelopmentandCooperationPITPPMLoissonMaximumLikelihoodQMTTQualifiedDomesticMinimumTop-upTaxTCJATaxCutsandJobsActof2017VATValue-AddedTax71IntroductionInternationalspilloversincorporatetaxationhavebecomeaprominentpolicyconcerninrecentyears,reflectingincreasedrecognitionoftheimpactthateachcountry’staxpoliciesmayhaveonbothactivityandtaxrevenuesinothers,andofthepotentialthiscreatesforpotentiallydamagingtaxcompetition.TheseconcernspromptedtheG20/OECD-ledprojectonBaseErosionandProfitShifting(BEPS)whichhasnowculminatedinthegenuinelyhistoricagreementamongnearly140countries,inOctober2021,onfundamentalreformoftheinternationaltaxarchitecture.1Akeyandunprecedentedelementofthisisagreementonaglobalminimumeffectiverateofcorporatetaxof15percent,duetotakeeffectin2024.Onemighthaveexpectedthesepolicydebatestohavebeeninformedby,andgenerate,closeanalysesoftheeffectsofinternationaltaxreformoncross-borderinvestment.Afterall,attractinginwardinvestment,includingthroughtaxsystemdesign,isevidentlyamajordgrowthenhancingknowledgespillovers.2Andthereisindeedalargeliteratureonthetopic:largeenoughtohavebegungeneratingmeta-studiesmanyyearsago(DeMooijandEderveen(2003),FeldbothacademicandintheOECD/G20discussions,towardsanalyzingtheanatomyand,inparticular,thetaxrevenueconsequencesof‘profitshifting’(meaningartificialtransactionsintendedsimplytoreducetotaltaxliability).3EmpiricalunderstandingoftaxeffectsonntpreoccupationwiththerevenueconsequencesofprofitshiftingalsostressedbySu′arezSerrato(2019),isindeedlikelyonereasonforthis.Butthereisperhapsamore1OECD(2020a,b).2Thereisampleevidenceonthepositivespillovereffectsfromforeigndirectinvestmentontheaccumu-lationofknow-how(seeforinstanceBaldwin,BraconierandForslid(2005)andKeller(2010)),economicgrowthforcapital-importingcountries(BosworthandCollins(1999),Javorcik(2004),andAlfaroetal.(2004)),anddomesticcapitalformationincapital-exportingcountries(Desai,FoleyandJr.(2005)).3RecentreviewsareinBeer,DeMooijandLiu(2020)andDharmapala(2019).Theexactmagnitudeofprofitshiftingremainscontentious,buthasbeenestimatedasdissipating5-10percentoftotalcorporateincometaxrevenueinadvancedandemergingeconomies(OECD(2015)andT?rsl?v,WierandZucman(2018)),athirdormoreofallU.Scorporatetaxrevenue(Clausing(2020)),andupto1.3percentofGDPindevelopingcountries(Crivelli,DeMooijandKeen(2016)).8fundamentalexplanationthanshiftinginterests.ThisisthatagreatdealoftheliteraturehasusedForeignDirectInvestment(FDI)data,enthusiasmforwhichhasbeendimmedbytheincreasingrecognitionofapointfirstmadelongago:“FDIdoesnotcorresponddirectlytoanymeasureofrealinvestment[...]Itismoreaccuratelythoughtofasameasureoffinancialflows...”4MeasuredFDIincludes,importantly,flowsthatfortaxreasons(routingintra-grouppaymentssoastotakeadvantageofreducedwithholdingtaxratesandothertaxortreatyprovisions),andperhapsforotherreasonstoo,passthroughsome‘conduit’jurisdictionwhilegeneratingrealinvestmentsomewhereelse.Thismeansthatthesameunderlyinginvestmentcanberecordedmultipletimesasitpassesthroughconduits,givingrisetoaformofdoublecounting.AndsuchconduitflowscannotbeunravelledfromtheFDIdata,whichtypicallyreportthelocationofaffiliates’immediateparents,notthatoftheultimateparentcompany.Theimportanceofthesepointsjumpsoutfromthedata,intheexistenceof‘investmenthubs’withinwardinvestmentclearlyfarinexcessoflevelsthatareplausiblefordomesticinvestment,giventhesizeofthedomesticeconomy.5Suchinvestments,withnoreallinkstothelocaleconomy,mayaccountforaround40percentofglobalFDI.6RelyingonFDIdatatoinfertaxeffectsoncross-borderrealinvestmentisthus,atbest,highlyproblematic.ThereareofcourseanalysesthatdonotuseaggregateFDIdata,withmuchattentionWhilethishasevidentimportanceandappeal,resultsfromaggregateFDI-basedstudiescontinuetoshapeviewsontax-responsiveness;anduntanglingwhatFDIdatadoordonotsayabouttaxeffectshasitsownimportancefromthewidermacroeconomicperspectiveofunderstandingandmanagingcross-bordercapitalflows.Moreover—evenleavingasidetheproblemswithFDIdata—muchoftheliteraturefocusesonlyontheimpactoninwardFDIofhostcountrytaxation,andsosimplydoesnotspeaktothekeyissueofcross-country4Slemrod(1990),p.83.5SeeforinstanceInternationalMonetaryFund(2014),BlanchardandAcalin(2021),Damgaard,ElkjaerandJohannesen(2019)andCoppolaetal.(2021).6Aykut,SanghiandKosmidou(2017),Damgaard,ElkjaerandJohannesen(2019).9spilloversfromonecountry’staxsystemtorealinvestmentelsewhere.7onwhatarrangements:theimpactonrealinvestment.Theresultsthenenableustoaddressakeybutsomewhatneglectedaspectoftheprospectiveimpactoftheglobalminimumtax:itseffectsonrealinvestment,bothbycountryandinaggregate.anewly-constructeddatasetonForeignAffiliateInvestment(FAI).Thesedataexplicitlyrecordtheacquisitionoftangibleassets(newandold)byforeignaffiliatesineachhostcountry,8andseethroughconduitflowstoidentifythecountryoftheultimateparent.Consequently,neitherofthemainshortcomingsofFDIdataapplies.Estimationcanexploretheeffectsonhost-parentbilateralFAIflowsoftaxsystemsinthehostcountry,parentcountryandotherpotentialhostcountries.Ourdatasetcoversinwardinvestmentfrom187parentcountriesintoaffiliatesin32(advanced)hostcountriesfrom2003-2016,alongwithdetailonarangeofactivityvariables.Therichinformationinthisdatasethelpsdisentangletheeffectoftaxesfromunobserved,confoundingfactorsoninwardforeigninvestment.Theofintangiblesinmultinationalactivity;buttherecanbelittledoubtastotheimportanceandsalienceoftangibleinvestmentfortheformationoftaxandotherpolicies.fectinwardFAICandidatesexploredinpreviousworksometimesofDevereuxandGriffith(2003))andmarginaleffectivetaxratesofthekinddevelopedin7Thesemi-elasticityonwhichthemeta-analysesofDeMooijandEderveen(2003)andFeldandHeck-emeyer(2011)focusisthatwithrespecttothehostcountryrate.Somestudiesdoconstructsimpletaxdifferentials:B′enassy-Qu′er′e,Fontagn′eandLahr`eche-R′evil(2005),forexample,examinetheeffectoftaxdifferentialsonbilateralFDI,thoughthedifferentialisinthiscasesimplythatbetweenhostandparentstatutoryrates.8TheythusputtorestthelamentwithwhichthequotationfromSlemrod(1990)continues:“Unfortu-nately,nodataexistonrealinvestmentmadebyforeignbranchesandsubsidiaries.”9TheKing-Fullertonmarginaleffectivetaxrateisthewedgebetweenthepre-andpost-taxreturnsonananalyticalworkorpublicdebate.Thereducedmarginaleffectiveratesassociatedwiththe2017TaxCutsandJobsActintheUnitedStates,forexample,promptedconcernsinCanadawiththeimpactoninvestmentthere.Buthowexactlythemarginaleffectiverateinonecountry,constructedwithaclosedeconomyinmind,shouldimpactinvestmentinanotherisbynomeansobvious.Similardoubtsapplymoregenerallytothehelpfulnessofthemuch-notedcross-country‘leaguetables’ofmarginaleffectivetaxrates10inunderstandingcross-borderTheapproachtakenhereistogroundtheempiricsinasimplebutreasonablygeneralmodeloftheinvestmentdecisionsofamultinationalthatpotentiallyoperateinseveralcoun-tries.Thisgeneratesasufficientstatisticwhichsummarizestheimpactoftheinternationalanddomestictaxsystemoninwardinvestment,encapsulatingeffectsthroughstatutorytaxrates,inboththehostcountryandalternativelocationsabroad,andKing-Fullerton-typemarginaleffectivetaxrates.Indoingso,ithighlightsoneformofspilloverthathasbeenlargelyneglectedintheinternationaltaxliterature:thedirecteffectthatinvestmentinoneaffiliatewithinamultinationalmayhaveontheprofitabilityofothers.Itmightbe,forexample,thatinvestinginone,andexpandingitssales,reducesthepriceatwhichothersducetheprofitabilityofothers.Werefertothis,forwantofabetterterm,as‘implicit’profitshifting,anddistinguishitfrom‘paper’profitshiftingofthekindthat,asmentionedabove,hasbecometheprimaryfocusofattention.Intheempirics,weallowfor,andattempttodifferentiatebetween,thesetwoformsofprofitshiftingandthetaxeffectsthatoperatethroughthem.eworkthatarticulatesection3elaboratesonthefundamentaldifferencesbetweenFAIandFDIdataandSection4thenunderlyinginvestmentwhichjustyieldstheinvestortheirrequiredpost-taxreturn(conventionallyexpressedasaproportionoftheformer).10As,notably,Mintz(2018)andBazelandMintz(2020).andsomeextensions,andexploresdirectlytheempiricalconsequencesofusingFAIratherthanFDIdata.Section6appliestheresultstoassesstheimplicationsforrealinwardinvestmentoftheprospectiveglobalcorporateminimumtaxrate.Section7concludes.2ChannelsofTaxSpilloverTofixideas,consideramultinationalpotentiallyinvestingthroughcontrolledaffiliatesinNcountries.Supposetoo,fornow,thatprofitshiftingthroughartificialtransactionsbetweenthemisnotpossible:relaxingthiswillbeoneofthelaterextensions.DenotingbyKj≥0thetangiblecapitalthatthemultinationalinvestsinitsaffiliateinjurisdictionj,andbyFj(Kj)someoutputassociatedwiththatcapital,wesupposetaxablereceiptsoftheaffiliateinthetypicalhostjurisdictionhtobeoftheformRh[F1(K1),..,Fh(Kh),..,FN(KN)].11Thekeyfeaturethisallowsforisthepossibilitythatreceiptsofanaffiliatelocatedinanycountryhmaydependontheproductionofrelatedaffiliatesinothercountries.Thisservestocapturearangeofpossiblepatternsofmultinationalactivity.Thesimplestsuchpossibilityisthateachaffiliateservesitsownlocalandcompetitivemarket,sellingitsproductatpricePh,sothatRh=Ph.Fh(Kh);thisisineffectthefamiliarclosedeconomycase,withnocrosseffectsacrossthedistinctaffiliates.Buttherearemanyotherpossibilitiescapturedbythisstructure.One,acaseofparticularinterestinanalyzingtheconsequencesofclosereconomicintegration,isthattheaffiliatessellahomogeneousproductintoasingleintegratedmarketwithinwhichthemultinationalhassomedegreeofmarketpower,sothatRh=P(對Fi)Fh,whereP(.)denotesinversedemand,withP\<0.12Inthiscase,eachcrosseffect?Rj/?Fh?P\Fjisstrictlynegative:increasedotherthanhitself.Anotherpossibilityisthatexpandingoutputinaffiliatehdivertsscarcemayalsobecircumstances,however,inwhichcrosseffectsarepositive,reflectingsome11TheFj(Kj)areassumedincreasingandstrictlyconcave;intermediategoodssuppliedbythirdpartiesistakentobesubsumedintherevenuefunctions.12Derivativesareindicatedbyaprimeforafunctionofonevariable.formofcross-bordercomplementarinessacrosshostcountries.(Theideaofcross-countrycomplementaritieshasattractedsignificantattentionintheliterature,butwithafocusquitedifferentfromthathere:thereitisontherelationshipbetweendomesticinvestmentandinvestmentabroad;13hereitisontherelationshipsbetweeninwardinvestmentintoalternativelocations.)Intheanalysisthatfollows,whilethepossibilityisleftopenthatfordefiniteness,thattheaggregatecrosseffectoveraffiliatesisstrictlynegative,sothat對h?Rj/?Fh<0.Thecorporatetaxsystemineachcountryjhastwocomponents(weignorepersonalincometaxes).OnecapturesthemarginaleffectivetaxrateinthetraditionofKingandFullerton(1984);werefertothisastheKF-METR.Familiarfromtheanalysisofinvestmentinclosedeconomiesasasufficientstatisticfortaxeffectsoninvestment,thisismodeledastaxofMj(expressedasaproportionoftherequiredreturn,whichisdenotedbyρandtakentobefixed)appliedtotheuseofrealcapitalinj.Theotherisa‘source-based’statutorytaxrateontheprofitearnedoninvestmentsinj(againstwhichMjisdeductible)atrateTj.Themultinational’saggregateafter-taxprofitisthusΠ(K1,..,KN)=(1?Tj)(Rj[F1(K1),..,FN(KN)]?ρ(1+Mj)),(1)pitalislocated.Allthis,itshouldbenoted,relatestoamultinationalparentedinsomeparticularcountryp;thatcouldberecognizedbytheclutterofanadditionalsubscript,butthepointonlybecomesmaterialifresidence-basedtaxesapply,andtheimplicationsofthataremorereadilydevelopedwhenitcomestothelaterempirics.From(1),thenecessarycondition14onthemultinational’schoiceofinvestmentinthe13SeeforexampleBeckerandRiedel(2012),usingfirmleveldataforEUfirms,andDesai,FoleyandHines(2009)fortheUnitedStates.Notetoothat,beingrooteddirectlyincommercialrelationsbetweenaffiliates,thecomplementarinessinmindherediffersfromthatwhichcanariseindirectlyastaxchangesactonthecostofcapitalbyaffectingthegainsfrom‘paper’profitshifting—thatisthemechanismatworkinSu′arezSerrato(2019),forexample,andisaddressedinSection5below.14ItisassumedthatΠisconcaveovernon-negativeK≡{K1,...,Kh}.typicalhostcountryh,Kh,is(1?Th){F?ρ(1+Mh)}+F(1?Tj)≤0(2)withaninternalsolutionforKhifftheequalityholds.Thefirsttermhereisfamiliarfromtheclosedeconomysetting,beingthepost-taxmarginalrevenueproductofcapitalinhnetofthetax-inclusivecostofcapital.Iftherewerenosecondterm,thiswouldhavethefamiliarimplicationthattheKF-METRisasufficientstatisticfortaxeffectsoninvestmentinh,withthestatutoryrateitselfhavingnoindependentimpact.Itisthesecondtermwhichisthenoveltyintroducedbytherevenuestructureintroducedabove,capturingtheimpactontheincentivetoinvestinhoftheconsequentcrosseffectonthenetrevenuesofaffiliateselsewhere.If,forexample,thataggregatecross-effect(aftertax)isnegative,thenfollowingtheclosed-economyruleofequatingthemarginalrevenueproductofcapitaltotheKF-METR-inclusivecostsofcapitalwilllead(giventheassumednegativityoftheoverallcross-effect)totoohighalevelofinvestmentinh,becausedoingsoignorestheadverseimpactofinvestingthereontheprofitabilityofotheraffiliatesinthemultinationalgroup.Thekeypointisthatinterdependenceinthecircumstancesoftheaffiliateswithinthemultinationalgroupmeansthatoneaspectofinvestinginhisthatmayserveasan‘implicit’profitshiftingdevice,inthesenseofmovingtaxableprofitsacrosscountries.Throughthisroutetheattractionsofinvestinginhdepend(throughthetermsmultiplyingF)onstatutorytaxratesinthehostcountryrelativetothosehandelsewhere.Thishasevidentsimilaritieswiththemoretraditionalideaofwhatwerefertohereas‘paper’profitshiftingasamatteroftaxavoidancerealizedthroughartificialtransactions.Here,however,theprofitshiftingistieddirectlyto,andimplicitin,therealinvestmentdecisionsthatarethemultinationalscorebusiness.15Thewayinwhichtaxationconsequentlyaffectsthelevelsandallocationofinvestment15Oneotheraspectofthedifferencebetween‘implicit’and‘paper’profitshiftingasthetermsareusedhereisworthnoting:whilebothshiftprofitsinthesenseofincreasingtheminsomecountrieswhilereducingtheminothers,implicitprofitshiftingwillgenerallyaffectthemultinationalgroup’sconsolidatedprofit,whereastheessenceof‘paper’profitshiftingisthat(apartfromanycostassociatedwiththeartificialarrangementsthemselves)itdoesnot.acrossthemultinationalgroupmoregenerallycanbemostclearlyseenbyrecasting(2)intothefamiliarstructureofadirectcharacterizationoftheequilibriummarginalrevenueproductofcapitalinh,therelevantmarginalrevenuenowbeingthatofthegroup:1,...N,by?R?Fh.F(Kh)≤ρ(1+I-METRh?RI-METRh≡Mh??hI-METRh≡1+?h(δ,T),?h(δ,T)≡αh(),αh≡?(?Rj/?Fh)/(?R/?Fh)≥0zh?Ri/?Fh. zh?Ri/?Fh.δhj≡Proof:ThisisthespecialcaseofamoregeneralresultinAppendixAobtainedbysetting?hthere(definedafter(10)below)tozero.ThereisalottounpackinProposition1.16Itshighestlevelimplicationisthat,asthenotationisintendedtosuggest,theI-METRhcombinestheNstatutoryratesandtheusualclosedeconomymarginaleffectiverateinthehostcountryintowhatcanbethoughtofasageneralizationtoaninternationalsettingofthestandardKF-METR.Thatis,justasKF-METRhsummarizesallrelevanttaxeffectsforanenterpriseoperatinginjustonecountry,sothesetofallI-METRhissufficienttocharacterizetaxeffectsoninvestment16Mattersareevidentlymorecomplex,forinstance,thanthesimpleview,oftenheardinpolicycontexts,thatmultinationals’locationdecisionsaredrivenbycross-countrycomparisonsofsomekindofaverageeffectivetaxrate,alongthelinesofDevereuxandGriffith(2003),whilethelevelofinvestment,conditionalonlocation,thendependsontheKF-METRinthehostcountry.Thedifficultywiththatintuitionisthatthefirstpartcomesfromviewinginvestmentasfixedinscale,whilethelatterpresumesittohavevariableinallcountriesinwhichthemultinationalmightoperate.Thisissointermsofboththeintensivemargin(thelevelofinvestmentinhostcountryh,conditionalonsomeinvestmenttakingplacethere)and,giventheassumedabsence—fornow—oflocationalfixedcosts,fortheextensivemargin(whethertoinvestinhatall).17ThecriticaldifferencebetweentheI-METRandthefamiliarKF-METRisthattheformerallowsforinteractionsinproductionandrevenuesacrosstheaffiliateswithinamultinationalgroup:18iftherearenone,theI-METRreducestothestandardKF-METR.19Moreprecisely,thestructureofI-METRhshowsthattaxationaffectsthemultinational’sinvestmentinhthroughtwochannels.Thefirstisthroughtheterm?h(δ,T),whichwerefertoasthe‘(statutory)ratedifferential’.This,inturn,servesasasufficientstatisticfortheimpactofstatutorytaxrates,inalljurisdictions,oncross-borderinvestmentdecisions.And,beneaththesomewhatheavynotation,ithasastraightforwardstructure.Inthenumeratorof?histheexcessofaweightedaverage20ofthestatutorytaxratesincountriesjotherthanthehosthoverthatinthehostitself,withtheweightδhjattachedtothestatutorytaxrateinjbeingtheproportionoftheadditionalrevenuetotherestofthemultinationalgroupthatisgeneratedbyincreasedinvestmentinhwhichaccruestotheaffiliateinj.Itisthisratedifferentialtermthatcapturestheimplicitprofitshiftinginherentinthecross-countryallocationofinvestment.Through?h,anincreaseinthetaxrateabroad,insomejh,tends21toincreaseinvestmentinthehostcountryhif,andonlyif,?Rj/?Fh<0,sothatthecrosseffectisnegative:intuitively,anincreaseinTjthenmakesitmoreattractivetoinvestinhbecausebecomemoreheavilytaxedComple17Howthelatterpartofthisconclusionisaffectedbythepresenceoflocationalfixedcostsistakenupinction18Afurtherdifference,ofcourse,isthatwhiletheKF-METRdependsonlyontaxparameterstheI-METRalsodependsonthestructureofthemultinational’soperations.19Becausethenαh=0.20Weusethistermforbrevity,thoughitshouldbenoted,however,thatsinceitmaythat?Rj/?Fh>0forsomejh(eventhough,byassumption,thiscannotbesointheaggregate),theweightsmaynotallliebetweenzeroandone.21Intheinformalcomparativestaticsthatfollowwesimplifybytreatingthevariousderivativesoftheaffiliates’revenuesasconstants.mentaritiesinproductionacrossthetwoaffiliates,ontheotherhand,wouldimplyreducedinvestmentinhwhenTjincreases.Anincreaseinthetaxrateinthehostcountryitself,Th,ontheotherhand,unambigu-ouslyreducesinvestmenttheredirectlythroughthenumeratorof?h.Thismaythoughbemitigatedbyaneffectthroughthedenominator,whicharises(recallingforinstanceequation(2))fromtheincreasedvalueofdeductingthefinancialcostsofinves