EUROPEANCENTRALBANK

EUROSYSTEM

OccasionalPaperSeries

IvelinaIlkova,LauraLebastard,RobertaSerafiniGeopoliticsandtradeintheeuro

areaandtheUnitedStates:

de-riskingofimportsupplies?

No359

Disclaimer:ThispublicationshouldnotbereportedasrepresentingtheviewsoftheEuropeanCentralBank(ECB).TheviewsexpressedarethoseoftheauthorsanddonotnecessarilyreflectthoseoftheECB.

ECBOccasionalPaperSeriesNo3591

Contents

Abstract2

Non-technicalsummary3

1Introduction4

1.1Contributionofthispaperandkeyfindings4

1.2Anoverviewoftheliterature6

2Evidenceoftradediversificationandrebalancing8

2.1Geopoliticaldimensionofdiversification8

Box1.Identifyingstrategicdependencies:adaptingtheEUapproachto

theUScontext9

Box2.Databaseandeventstudyspecifications13

2.2Limitedrebalancing:persistentrelianceongeopoliticallydistant

sourcingcountries15

Box3.Comparisonofthesourcesofstrategicgoodsimportedbythe

euroareaandtheUnitedStates17

3Currentandpotentialeffectsonprices21

3.1Thecostofdiversificationandrebalancing21

3.2Priceimplicationsofshiftingtowardsgeopoliticallyclose

sourcingcountries23

4Conclusions26

References28

Annex30

Acknowledgements36

ECBOccasionalPaperSeriesNo3592

Abstract

Basedongranulardataattheproductlevel,thispaperlooksatwhetherandhowtheeuroareaandtheUnitedStateshavemodifiedtheirimportsourcingstrategiessince2016,theroleplayedbygeopoliticaltensionsandthepotentialimpactonimport

prices.Itconsiderstwodifferent,butnotmutuallyexclusive,changestosourcing

strategiesforagivenproduct:(i)increasingthenumberofsourcingcountriesand

(ii)reducingtheimportmarketshareofthemainsuppliercountry.Datasuggestthatbothregionshave,onaverage,increasedthenumberofsourcingcountries,

particularlyforproductsthataremostlyimportedfrom“geopoliticallydistant”

countries(basedonUNGeneralAssemblyvotingrecords).Broadeningthenumberofsuppliercountrieshascomeatacost;however,ithasaffectedonlyasmallshareoftotalimports,withmodestimplicationsforinflationandthetermsoftrade.Atthe

sametime,evidenceofareductionintheimportshareofthemainsuppliercountry

ismoremixedandisgenerallyassociatedwithashifttowardscheaper–butnot

necessarilygeopoliticallycloser–countries,suggestingthatcostconsiderationstakeprecedenceoversupplychainresilienceandnationalsecurityconcerns.

Keywords:fragmentation,geopolitics,diversification,rebalancingJELcodes:F14,F51,F62

ECBOccasionalPaperSeriesNo3593

Non-technicalsummary

Tradetensions,theCOVID-19pandemicandthewarinUkrainehavehighlightedvulnerabilitiesinforeignsourcingofintermediateandfinalgoods,heightening

appreciationofgeopoliticalrisks.Thisoccasionalpaperlooksathowthatisaffectingtradepatterns.Itusesdetailedproduct-levelinformationinordertounderstandhowUSandeuroareafirms’sourcingstrategieshavechangedsince2016andascertainthepotentialcostsofthosechanges.Itfocusesontwonon-mutuallyexclusive

strategiesaimedatloweringrisksofsupplychaindisruptionforagivenproduct:

(i)diversificationofsourcingcountries(extensivemargin)and(ii)rebalancingby

reducingthemarketshareofthemainsuppliercountry(intensivemargin).Wethenassesswhetherthoseshiftshaveageopoliticaldimension.

Wefindthatbothregionshavediversified,increasingthenumberofsourcing

countriesonaverageforeachproduct.Thathasacceleratedsincethepandemic,

especiallyintheeuroarea.Geopoliticalriskshaveplayedarolehere:diversificationhasbeenstrongerforproductswheretheeuroareaandtheUnitedStatesusedtorelyparticularlyheavilyonimportsfromgeopoliticallydistantcountries(e.g.China,RussiaandIran).Diversificationhascomeatacost,asthenewsourcingcountriesare,onaverage,moreexpensivethantheprevioussuppliers.However,theimpactonaggregatepriceshasbeensmall:overtheperiod2016-23,productsfromnew

countriesaccountedforaverysmallshareoftotalimports(0.2-0.3%),implyingmodesteffectsoninflationandthetermsoftrade.

Furthermore,whiletherehavebeenstrongdeclinesinChina’sshareoftotalUSimportssince2018andinRussia’sshareoftotaleuroareaimportssince2022,evidenceofrebalancingawayfromothergeopoliticallydistantcountries(i.e.

reductionsinthemarketsharesofsuchsuppliercountries)islessclear.Ingeneral,

geopoliticallydistantcountries’sharesoftotaleuroareaandUSimportsremainhigh.

Indeed,China’sshareoftotaleuroareaimportshasincreasedby3percentage

points(ariseof15%)since2016.ForthoseproductswheretheUnitedStatesandtheeuroareahavereducedtheirrelianceonamainsupplier,rebalancingseemstohaveprimarilyreflectedcostconcerns–i.e.adesiretoshifttowardscheaper

sourcingcountries–ratherthansupplychainresilienceornationalsecurity

considerations.Indeed,evidencesuggeststhatrebalancingtowardsgeopoliticallyclosecountriesiscostly,implyingashifttowardsimportsthatare20-40%more

expensiveonaverage.

ECBOccasionalPaperSeriesNo3594

1Introduction

Inrecentyears,aseriesofadverseshocks–US-Chinatradetensions,the

pandemicandthewarinUkraine–havehighlightedvulnerabilitiesrelatedto

foreignsourcingofintermediateandfinalgoodsandintensifiedtheriskof

geoeconomicfragmentation(i.e.thecreationofeconomicblocsalonggeopoliticallines).Beingmoreoutward-orientedthantheUnitedStatesandChina,theeuroareamaybeparticularlyexposedtosuchfragmentationrisks(Babaetal.,2023).

Understandinghowsourcingischangingmattersbecause,assurveysshow,firmscanadoptdifferentstrategiestoaddressfragmentationrisks,with

differenteconomicoutcomes.Firmscanusethreebroadstrategiestoadjusttheirsourcing.First,theycandiversifysourcingacrossmorecountries.Second,theycanrebalancetheirforeignexposure(i.e.reducetheirdependenceononeortwomain

countries,reducingthosecountries’sharesoftotalimports),possiblyshiftingtowardscountriesthataregeographicallycloser(near-shoring)orgeopoliticallycloser

(friend-shoring).Andthird,theycanswitchtodomesticsuppliers,leadingtoarelocationofproductioninthehomecountry(re-shoring).1

Differentde-riskingstrategiescanhavedifferentimplicationsforglobalandeuroareatrade.2Insomecases,theycouldleadtofurtherglobalisationora

reconfigurationofsourcingpatterns(potentiallyresultinginmorelocalisedor

regionallyfocusedsupplynetworks).Alternatively,theycouldresultindeglobalisationifeconomiesbecomemoreinward-looking.Domesticandforeignproductionand

investmentcouldbesignificantlyaffected,andpricingdynamicscouldbealtered.Obtainingpromptinsightsintoongoingdevelopmentsisthereforeusefulinordertoanticipatepotentialimplications.

1.1Contributionofthispaperandkeyfindings

ThispaperlooksathowtheeuroareaandtheUnitedStateshaveadjusted

theirsourcingstrategiessince2016andassessesthepotentialcostsofthoseshifts.First,itassesseswhethersourcingpatternshavebeenadjustedandinwhichdirection–i.e.itlooksatwhethercountrieshavediversifiedtheirimports(increasingtherangeofcountriestheyimportfrom)and/orrebalancedtheirimports(reducing

themarketshareofthemainsuppliercountry).3Second,itassesseswhethersuch

1Inthesamewayassourcing,productionactivitycanalsobeshiftedtogeographicallyorgeopoliticallyclosercountries,movedclosertoacountry’smainsalesmarketsorevenshiftedtothehomemarket.

Furthermore,firmsmaydecidetoadjustbusinessprocesses,bymovingfroma“justintime”toa“justincase”modelandholdinglargerinventories.Whiletheseaspectsarepotentiallyrelevant,theycannotbeinvestigatedusingtradedata,sotheyliebeyondthescopeofouranalysishere.

2Inaddition,thesestrategiesalsodiffersignificantlyintermsofthecostsandtimerequired(CroweandRawdanowicz,2023).

3Withourdata,itisnotpossibletotellwhethergoodscomefromanindependentsupplierorabranch(i.e.whetheritisthesourceortheplaceofproductionthathaschanged).

ECBOccasionalPaperSeriesNo3595

shiftsintradepatternshaveageopoliticaldimension.Finally,thepaperquantifiestheimpactthatsuchreorganisationofsupplychainsmayhaveonimportprices.

Theassessmentfocusesontwopossiblestrategies:diversificationand

rebalancing.Throughouttheanalysis,itshouldalwaysbeborneinmindthat

diversificationandrebalancingmayoccureitherseparatelyoratthesametime.

Chart1

providesavisualexampleofhowthetwoconceptsrelatetoeachother,withthestartingpointbeingascenariowhereacountryimportsagivenproductfrom

countriesAandB(thelatterbeingthemainsupplier).Sourcingcanbecomemore

diversified,movingfromtwotofoursourcingcountries,withoutchangingthemarketshareofthemainsupplier(panela);inthesamevein,rebalancingcanoccurwithoutincreasingthenumberofsuppliers(panelb).4

Chart1

Diversificationvsrebalancing–avisualexample

(percentagesoftotalimports)

A40%

B60%

a.diversificationb.rebalancingc.diversification&rebalancing

C1%

B60%

C1%

D1%

D1%

B51%

A49%

A38%

B51%

A47%

Source:ECB.

Thisassessmentfocusesondevelopmentssince2016usingdetailed

product-leveldata.Wefocusondevelopmentssince2016becausethisappearstobethepointatwhichgeopoliticalconsiderationsbegantoplayastrongerrolein

traderelations,withtheelectionofDonaldTrumpandthebeginningoftrade

tensionsbetweentheUnitedStatesandChina.Theanalysisisbasedonproduct

dataatthesix-digitleveloftheWorldCustomsOrganization’sHarmonisedSystem(HS)classification,correspondingtoverydetailedproductcharacteristics.5Thishighlevelofgranularityallowsustotrackthenumberofcountriesfromwhicheconomiesimporteachproductandthepricescharged.Anotheradvantageofusing

4Thispaperdoesnotseektoquantifythepotentialbenefitsofsuchde-riskingstrategies.Suchbenefitsarelikelytocomeintheformofenhancedresilienceandlowerwelfarecostsintheeventoffuture

geopoliticalshocks.Benefitscanalsostemfromareducedriskofthehomecountrybeingblackmailedbyageopoliticallypowerfulsourcingcountry.Reducingstrategicdependenceonpotentiallyunfriendlyautocraticstatesisaninsurancepolicythatenhanceswelfareifdoneefficiently,asitreducesthe

expectedcostsofgeopoliticalshocks.

5TheHS6classificationallowscross-countrycomparisonatahighlygranularlevel.Forinstance,

code040110correspondsto“dairyproduce;milkandcream,notconcentrated,notcontainingaddedsugarorothersweeteningmatter,ofafatcontentnotexceeding1%(byweight)”.

ECBOccasionalPaperSeriesNo3596

product-leveldataistheirtimeliness.Forexample,product-leveldataareavailablewithathree-monthlag,whilefirm-leveldatatypicallyhaveatwo-yearlag.6

Ourpapercontributestothefast-growingbodyofempiricalliteraturethatis

usinggranularproduct-leveldatatoexplorewhetherseeminglystable

aggregatetrendsaremaskingaredirectionoftradeflowsbetweencountries,potentiallysignallinggeopoliticalfragmentation(AlfaroandChor,2023;and

Freundetal.,2023).Tothebestofourknowledge,ourpaperisoneofthefirstto

provideahighlydisaggregatedcomparativeanalysisoftheeuroareaandtheUnitedStateswithemphasisonthepriceimplicationsoffragmentation–anaspectthathasbeenrelativelyunderexploredinpreviousresearch.

Wefindevidenceofdiversification,whichhasincreasinglyhadageopoliticaldimension.TheeuroareaandtheUnitedStateshavebothdiversified,increasingthenumberofsourcingcountriesforeachproductonaverage.Geopoliticalrisks

haveplayedarolehere:diversificationhasbeenstrongerforproductswheretheeuroareaandtheUnitedStatesusedtorelyparticularlyheavilyonimportsfromgeopoliticallydistantcountries.Diversificationhascomeatacost,asthenew

sourcingcountriesare,onaverage,moreexpensivethantheprevioussuppliers.

Incontrast,evidenceofrebalancingismorelimited.AggregatedatashowstrongdeclinesinimportsfromChinatotheUnitedStatespost-2018andimportsfrom

Russiatotheeuroareapost-2022.However,evidenceofrebalancingawayfrom

othergeopoliticallydistantcountriesislessclear.ForthoseproductswheretheeuroareaandtheUnitedStateshaverebalanced,thisseemstohaveprimarilyreflectedcostconcerns–i.e.shiftstowardscheapersourcingcountries–ratherthansupplychainresilienceornationalsecurityconsiderations.

1.2Anoverviewoftheliterature

Geopoliticaltensionshavedominatednewspaperheadlinesandprompted

numerousadhocsurveys,indicatingthatfirms’sourcingstrategiesareindeedgraduallyshifting.Whenaskedaboutpotentialchangestotheirsourcing

strategies,halfofallfirmsrespondingtotheEIB’s2023InvestmentSurvey(bothEUandUSfirms)havechangedorareplanningtochangetheirsourcingstrategyby

increasingthenumberofcountriestheyimportfromandincreasingtheirinventories(EuropeanInvestmentBank,2023).ThetendencytowardsgreaterdiversificationofinputsourcingisalsoconfirmedbyarecentECBsurveyofleadingfirmsoperatingintheeuroarea,withdatashowingthatgeographicallyandgeopoliticallycloser

countriesarebeingfavoured(Attinasietal.,2023b).NationalsurveysongeopoliticalriskscoordinatedbytheBancad’Italia,theBancodeEspa?aandtheDeutsche

Bundesbankhavealsoincludedsmallfirms(Balteanuetal.,2024).ThesesurveysrevealthatasignificantpercentageofcompanieshavetakenstepstomitigatetheirsourcingriskfromChina–40%inGermany,30%inItalyand22%inSpain.

6However,thegranularproduct-leveldataonlygiveusabroadpictureanddonottelluswhetherparticulartypesoffirm–suchasmultinationalsorsmallandmedium-sizedenterprises–pursueparticularsourcingstrategies.

ECBOccasionalPaperSeriesNo3597

Additionally,between20%and27%offirmsinthesecountriesareconsideringtakingsimilaractionbytheendof2024.

Empiricalanalysis,however,yieldsmixedevidencedependingontheregion

thatisbeinglookedat.Availableempiricalliteratureonfragmentationmostly

focusesontheUSeconomy,lookingatthetraderestrictionsthatwereintroduced

duringtheTrumpadministration.Freundetal.(2023),forexample,findsignificant

reshapingofUSglobalsupplychainsowingtotariffsimposedonChineseimports

since2018,withdeclinesinChina’ssharesoftotalUSimportsforthesetofproductsthatweresubjecttotheadditionaltax.Instead,firmshaveincreasedtheirimports

fromlargedevelopingeconomieswhicharefirmlyembeddedinChinesesupply

chains,particularlyforstrategicgoods.CountriessuchasVietnam,Thailand,SouthKoreaandMexicohaveemergedasmajorwinnersinglobalexportmarkets,partlybyprovidingsubstitutesforproductssubjecttotheUS-Chinatariffs(Fajgelbaumetal.,2024;andAlfaroandChor,2023).

Ontheeuroareaside,tradedatahavenotyetshownanysignificant

deglobalisationofproductionchains.Aggregateindicatorsofglobalvaluechainssuggestthatnoreconfigurationoftheeuroareaproductionchainhasyettaken

place,asnonoticeableshiftinthesourcingofintermediategoodshasbeen

observed(DiSanoetal.,2023).Thisissupportedbyfirm-leveldata:DiStefanoetal.(2022)foundthatthepandemicdidnottriggersignificantreshoringamongItalian

multinationalfirms.Similarly,deLucioetal.(2023),usingarepresentativesampleofSpanishmanufacturersinvolvedinglobalvaluechains,foundnoevidenceofashifttowardssourcingfromcountriesthatweregeographicallyorgeopoliticallyclosertoSpain.

Anumberofstudiesestimatethepossibletradeandwelfareconsequencesoftheworldfragmentingintoblocs.WithafocusontheEU,Attinasietal.(2023a)

quantifytheeconomiccostsofhypotheticalfragmentationscenariosusingthe

multi-country,multi-sectormodeldevisedbyBaqaeeandFahri(2023).Theyfindthatadecouplingoftheglobaleconomyintoawesternblocandaneasternbloc(broadly

comprisingadvancedanddevelopingeconomiesrespectively)wouldreduceEU

outputandtrade,aswellasraisingprices,whilewelfarelosseswouldgenerallybemoremuted.Camposetal.(2023)estimatethattheworldfragmentingintothree

tradeblocs(western,easternandneutral)wouldhaveimportanteffectsontradebetweenthem,reducingtradeflowsby22-57%inthemostextremescenarios

(involvingthewithdrawaloftheeasternblocfromtheWorldTradeOrganization

(WTO)).Althoughsmallerthantradelosses,welfarelosseswouldstillbesizeable,withthelargestlossesseenintheeasternbloc.GoesandBekkers(2022)and

Felbermayretal.(2023)find,usinggeneralequilibriummodels,thattheglobaleconomydecouplingintotwoblocsresultsinsignificantwelfarelosses.

ECBOccasionalPaperSeriesNo3598

2Evidenceoftradediversificationandrebalancing

2.1Geopoliticaldimensionofdiversification

Theeuroareahasprogressivelydiversifiedimportsourcesoverthepasttwodecades,althoughthereisnosignthatthisprocesshasaccelerated

significantlyrelativetothepast.SincetheelectionofDonaldTrumpasUS

Presidentin2016,diversificationintheeuroareahascontinued,withagradual

increaseinthenumberofsourcingcountriesperproductandaslightacceleration

sincethepandemic

(Chart2)

;thisincreasecanalsobeobservedforgoodsof

strategicimportance.7,8However,thisrepresentsthecontinuationofaprocessthat

hasbeenongoingforsomeyearsnow.Incontrast,therehasbarelybeenany

diversificationintheUnitedStates.Interestingly,in2023theeuroareahad,on

average,two-thirdsmoresourcingcountriesperproductthantheUnitedStates,and20%moreforstrategicgoods.Asshownin

ChartA2,

diversificationappearstohaveincreasedmorestronglyintheeuroareathanintheUnitedStatessince2016,evenaftercontrollingfortheinitialnumberofsourcingcountriesperproduct.

7See

ChartA1

intheannex.StrategicgoodsareasdefinedinthelistinEuropeanCommission(2021).TheEuropeanCommissionidentifiedstrategicdependenciesrelatedtospecificimportedinputs“inthemostsensitiveecosystemswheretheEUcanbeconsideredhighlydependentonimportsfromthird

countries”,basedonthreeindicators:(i)concentration,asmeasuredbytheHerfindahl-Hirschman

Indexandthemarketshareofnon-EUsupplyingcountries;(ii)theimportanceofdemand,calculatedasextra-EUimports’shareoftotalEUimports;and(iii)substitutability,calculatedastheratioof

extra-EUimportstototalEUexports.FortheUnitedStates,weconstructasimilarsetofproducts,adaptingtheEuropeanCommission’smethodologytofitUSdata(seeBox1fordetails).More

descriptivestatisticsonstrategicgoods(geographicaldistributionandconcentrationbysourcingcountry)canbefoundinBox3forboththeeuroareaandtheUnitedStates.

8PolicyinitiativessuchastheCHIPSActandtheInflationReductionActintheUnitedStatesandtheEuropeanChipsAct,theNet-ZeroIndustryActandtheCriticalRawMaterialsActinEuropehaveacommongoalofreducingdependenceoncriticalproductsandinputs.

ECBOccasionalPaperSeriesNo3599

Chart2

Numberofsourcingcountriesperproduct

(averagesandmedians)

50

45

40

35

30

25

20

15

10

5

0

Average–euroareaMedian–euroarea

Average–UnitedStatesMedian–UnitedStates

20022005200820112014201720202023

Source:TradeDataMonitor.

Notes:ProductsaredefinedatHS6level.Datafortheeuroarearefertoextra-euroareaimports.

Box1.Identifyingstrategicdependencies:adaptingtheEUapproachtotheUScontext

IvelinaIlkova

ThepandemicunderscoredthevulnerabilityoftheEUeconomytodisruptionsinthesupply

ofspecificproductsandinputs.Thisawarenessledtotheemergenceoftheconceptof“strategicgoods”–criticalforeigninputsessentialforEUproduction.Inresponse,theEuropeanCommissiondevelopedalistofstrategicgoods/dependencies,basedoncriteriasuchashighEUdemand,

relianceonconcentratedsourcesandlimitedsubstitutability.NosuchlistiscurrentlyavailablefortheUnitedStates.Thisboxexplainshowweaddressthisgap,outliningthemethodologyused,

discussingitslimitations,andproposingadaptationstosuitboththeUSandtheEUcontexts.

InthestaffworkingdocumentEuropeanCommission(2021),amethodologyforidentifyingstrategicdependenciesisdeveloped.Essentially,thatmethodologycombinesquantitativeand

qualitativeanalysiswiththeaimofcreatingalistofstrategicallyimportantproductswheretheEUisstronglydependentonforeignsuppliers.Themethodologycanbebroadlydividedintotwosteps.

First,thelistofHSproductsisfiltered,reducingitfrom5,000toroughly390goods.Thisisdonebycomputingthefollowingthreecoredependencyindicators(CDIs),withproductsselectedbasedoncriticalvaluesspecifictoeachCDI:

?CDI1capturestheconcentrationofEUimportscomingfromnon-EUcountries.Foreachproduct,itiscalculatedusingtheHerfindahl-HirschmanIndex,asdefinedbelow:

n

i=1

wheresiisthemarketshareofthenon-EUcountryandnisthetotalnumberofsourcecountries.Avalueof0.5meanstheproductissuppliedbytwocountries.Thehigherthevalue,thefewercountriessupplythisproducttotheEU.ThecriticalvalueabovewhichtheconcentrationisconsideredtoohighisCDI1>0.4.

ECBOccasionalPaperSeriesNo35910

?CDI2aimstomeasuretheimportanceofextra-EUimportsintotaldemand.Inotherwords,itidentifiesproductswhicharemainlysuppliedbyforeignsources,insteadof

locallythroughintra-EUtrade.Thecriticalvalueis0.5andtheformulausedtocalculateitisasfollows:

?CDI3isusedtomeasurethesubstitutabilityofextra-EUimportswithEU

production.ThisratioseekstomeasuretheextenttowhichEUproductioncancoverimportneedsintheeventofdisruption.EUexportsareusedasaproxyforEU

production,andthecriticalvalueis1:

CDI?=

Inasecondstep,thelistisnarroweddownfurtherbyselectinggoodsusedin“themost

sensitiveindustrialecosystems”,suchasdigital,health,aerospaceanddefence,andelectronicssectors.Thestaffworkingdocumentdoesnotprovideanexplicitlistofthesectorsthatare

consideredpartofthesesensitiveecosystems,makingitdifficulttofullyreplicatetheresults.

However,theEuropeanCommissionrecentlypublishedalistofCriticalTechnologyAreas,9whichweusedtobuildalistoftheHS2sectorsthatarepartofsensitiveecosystems(withtheexactHS2codesbeinginferred).

TheabovemethodologycannotbeappliedinfulltotheUnitedStates.Therearetworeasonsforthis.First,CDI2–whichmeasurestheimportanceofdemand–requiresdataonbothintra-and

extra-UStrade.Owingtotheunavailabilityofdataonintra-UStradeflows,CDI2cannotbe

calculatedfortheUnitedStatesunlessareasonableproxyfordomesticdemandattheHS6levelcanbefound.Second,theidentificationofsensitiveecosystemsintheEuropeanCommission’s

workingdocumentwasdonespecificallyfortheE