Globalization and Emerging Markets With or Without Crash

GlobalizationandEmergingMarkets:WithorWithoutCrash?

ByPHILIPPEMARTINANDH´EL`ENEREY*

Weanalyzetheeffectsoffinancialandtradeglobalizationonthelikelihoodof

financialcrashesinemergingmarkets.Whiletradeglobalizationalwaysmakes

crasheslesslikely,financialglobalizationmaymakethemmorelikely,especially

whentradecostsarehigh.Pessimisticexpectationscanbeself-fulfillingandleadto

acollapseindemandforgoodsandassets.Suchacrashcomeswithacurrent

accountreversalanddropsinincomeandinvestment.Lower-incomecountriesare

morepronetosuchdemand-basedfinancialcrises.Aquantitativeevaluationshows

ourmodelisconsistentwiththemainstylizedfactsoffinancialcrashesinemerging

markets.(JELF12,F32,F37,F41,O16)

Doemergingmarketsreapthebenefitsof

financialglobalization,enjoyingincreasedin-

vestmentandabetterabilitytodiversifyrisk?

Ordotheyfaceahigherlikelihoodoffinancial

crashasmorecapitalflowsin?Theempirical

literaturesupportsbothpossibilities.Ontheone

hand,anumberofpapersinfinanceshowthat

financialopeninginemergingmarketsleadsto

adecreaseinthecostofequitycapitalandcan

haveapositiveeffectondomesticinvestment.1

Ontheotherhand,avoluminousliteraturesur-

veyedbyJoshuaAizenman(2004)emphasizes

therisksofliberalizationandthevulnerabilityofemergingmarketfinancialsystemstocapital

mobility.CharlesWyplosz(2001)findsthatex-

ternalfinancialliberalizationisconsiderably

moredestabilizingindevelopingcountriesthan

indevelopedeconomies.GracielaKaminski

andSergioSchmukler(2001)showthatstock

marketsbecomemorevolatileinthethreeyears

followingfinancialliberalizationbutstabilizein

thelongerrun.

Interestingly,recentempiricalworkshows

thatgoodstradeopennessalsoinfluencesthe

frequencyofcrashesinemergingmarkets,but

intheoppositedirectiontofinancialopenness.

EduardoA.CavalloandJeffreyA.Frankel

(2004)findthattradeopenness(instrumented

bygravityvariables)reducesthevulnerability

ofcountriestosuddenstops.TheArgentinaof

the1990sisoftenpresentedasatypicalexam-

pleofafinanciallyopeneconomyrelatively

closedtogoodstrade.Ithassufferedheavily

fromsuddenstops(seeGuillermoA.Calvoet

al.,2003;CalvoandErnestoTalvi,2004).

Thesecontradictoryeffectsoffinancialand

tradeglobalizationsareillustratedinTable1.

Wereporttheaveragenumberoffinancial

crashesperyearfordevelopedandemerging

economies,dividingeachgroupalongthedi-

mensionsoffinancialandtradeopenness.2*Martin:UniversityofParis1Panthe´onSorbonneEconomie,ParisSchoolofEconomics,106-112bddel’Hoˆpital,75647ParisCedex13,France,andCentreforEconomicPolicyResearch(CEPR)(e-mail:philippe.martin@univ.paris1.fr);Rey:DepartmentofEconomicsandWoodrowWilsonSchool,307FisherHall,PrincetonUni-versity,Princeton,NJ08544,CEPR,andNationalBureauofEconomicResearch(e-mail:hrey@princeton.edu).Wethanktwoanonymousrefereesforveryhelpfulcommentsonapreviousversion,DanielCohen,Pierre-OlivierGourin-chas,GeneGrossman,GalinaHale,OlivierJeanne,EnriqueMendoza,RichardPortes,LarsSvensson,AaronTornell,aswellasparticipantsatmanyseminars.WealsothankGra-cielaKaminskyandSergioSchmuklerforthestockmarketdata.RachelPolimeniprovidedexcellentresearchassis-tance.ThispaperispartofaResearchTrainingNetworkon“TheAnalysisofInternationalCapitalMarkets:Under-standingEurope’sRoleintheGlobalEconomy,”fundedbytheEuropeanCommission(ContractNo.HPRN-CT-1999-0067).1See,forexample,GeertBekaertetal.(2005),PeterBlairHenry(2000),andAnushaChariandHenry(2002).Themacroeconomicliteraturefindsmoretenuousevidencethatfinancialopeningcontributestolong-termgrowth.SeeSebastianEdwards(2001),forexample.2Moreprecisely,emergingmarketsaredefinedasthosewithGDPpercapitaequalorbelowthatofSouthKorea.Thesamplecoverageforthosecountriesstartsattheearliestin1975andendsin2001.Acrashisdefinedasamonthlydropinthestockindex(indollars)largerthantwostandarddeviationsoftheaveragemonthlychange.Wedividedthesampleintoperiodsforwhichcountrieswerefinancially

1631Table1suggeststhatopeningtocapitalmove-

mentsisverypositivelycorrelatedwiththefre-

quencyofcrashesforemergingmarkets,butnot

forindustrializedcountries.Tradeopenness

(whethermeasuredbythetrade-to-GDPratioor

followingJeffreySachsandAndrewWarner,

1995),however,isassociatedwithalargede-

creaseinthefrequencyofcrashesforemerging

markets.3Hence,accordingtoTable1,beingan

emergingmarketopentofinancialflowswhile

closedtogoodsflowsmaximizesthefrequency

ofcrashes.

Thecontributionofourpaperistopresenta

generalframeworkinwhichthesecontradictory

effectsoffinancialandtradeliberalizationscan

bereconciled.Wecanalsomakesenseofthedifferentialimpactoffinancialglobalizationon

emergingmarketsanddevelopedeconomies.

Weemphasizethekeyroleofdemandand

marketsizeindrivingboththepositiveeffectof

financialintegrationonanemergingeconomy

anditsnegativeconsequences.

Inourmodel,theworldconsistsofone

emergingmarketandonedevelopedeconomy

whichdifferonlyintheirproductivitylevel.In

bothcountries,entrepreneursoperatinginmo-

nopolisticgoodsmarketsdecidewhethertofi-

nanceriskyfixed-sizedinvestments,sellshares

oftheseinvestmentsonthestockexchange,and

acquiresharesinotherriskyventuresdeveloped

athomeorabroad.Entrepreneursmayturnpes-

simisticandexpectlowlevelsofaggregatein-

vestment.Duetohomebiasingoodstrade,

negativeprospectsregardinginvestmenttrans-

lateintolowexpectedincomeanddemandfor

goods,lowprofits,andhencelowdemandfor

domesticassets.Thisvalidatestheirpessimistic

priorsanddetersthemfromdevelopingrisky

investments.Thehomebiasinfinancialmarkets

inturnimpliesthatthefallinincomeinthe

emergingmarketalsoleadstoafallindomestic

assetdemandandprices.Inthisequilibrium,

assetpricesandinvestmentcollapse,income

decreases,andacapitalflightoccurssincedo-

mesticagentsbuysharesinthedevelopedcoun-

trystockexchange.Thecircularcausalityis

magnifiediftradecostsarehigh,sincefirms’

profitsanddividendsinmoreclosedeconomies

aremoredependentontheleveloflocalde-

mand.Theyarethereforemoreatriskwhen

expectationsturnpessimistic.

Thelikelihoodofacrashishigheratan

intermediatedegreeoffinancialsegmentation.

Whenfinancialmarketsareperfectlyintegrated,

nofinancialhomebiasexistsandarbitrage

equatesassetprices,sothatlocalincomecon-

ditionsdonotalterthecostofcapitalin

theopenandfinanciallyclosed,followingKaminskyandSchmukler(2001).Hence,amongour62countries(34emergingcountries),31appeartwiceastheychangedstatusduringthesampleyears.Wealsoclassifiedcountriesintermsoftheiropennesstotrade.Wechosetwowidelyusedmeasuresoftradeopenness:(a)TheclassificationbySachsandWarner(1995)extendedtothe1990sbyRomainWac-ziargandKarenH.Welch(2003),whoprovideliberaliza-tiondatesforabroadsetofcountries.Thismeasureofopennessisbasedontradepolicies.Somecountrieshaveliberalizedfinancialandtradeflowsatdifferentdatesandhenceappearindifferentcellsofthetable.(b)TheaverageofexportsplusimportsoverGDPduringtheperiodcon-sidered.Thismeasureofopennesscapturesthedegreeofindependenceoftheeconomyonlocaldemand.Wecallopen(respectivelyclosed)acountrywhoseopennessratioisabove(respectivelybelow)themedianofitsgroup.Thetradeopennessratiocutoffforfinanciallyopenemergingcountriesis63percentofGDP.Forthegroupoffinanciallyclosedcountries,thetradeopennessratiocutoffis54per-cent.Formoredetails,seetheDataAppendixathttp://team.univ-paris1.fr/teamperso/martinp/table.pdf.3Thisisnotthecasefordevelopedcountries,forwhichthefrequencyofcrashesislowoverall.Fordevelopedeconomies,weusetheratiooftradetoGDPasourmeasureofopenness.AccordingtotheSachs-Warnermeasure,onlytwoindustrializedcountriesareclassifiedasclosedatsomepoint,sowedonotreporttheresult.TABLE1—FREQUENCYOFCRASHESANDOPENNESS

Tradeingoods

EmergingDeveloped

ClosedOpenClosedOpen

Financiallyclosed0.40a0.35b0.09a0.15b0.07b0.09a0.10bFinanciallyopen0.78a0.76b0.55a0.57b0.05b0.06a0.14b

aSachs-Warnermeasureofopenness.bTrade/GDPmeasureofopenness.1632THEAMERICANECONOMICREVIEWDECEMBER2006