Fiscal Policy, Hidden Deficits, and Currency Crise

31FiscalPolicy,HiddenDeficits,andCurrencyCrisesHomiKharasandDeepakMishraAbstractBudgetdeficits,asconventionallymeasured,sufferfrommanymeasurementandmethodologicalproblems.Researcherswhohaveusedtheseconventionaldeficitstoexplaincurrencycrises,therefore,unsurprisingly,havefoundlittleevidenceofanysystematiclinkbetweenthetwo.Weprovideanalternativedefinitionofbudgetdeficitandshowthatthereisacloseassociationbetweenthenumberofcurrencycrisesandourmeasureofdeficit.Ifaskedtopickasingleconceptthatbestmeasuresthefiscalsituationofaneconomy,manyofuswouldpick“budgetdeficit,”butbudgetdeficit,asconventionallymeasured,suffersfrommanyproblems.Theseproblems,whicharemostlymeasurement-relatedandmethodologicalinnatureandmoresevereindevelopingthaninindustrialcountries,arisebecauseofavarietyofreasons.Theserangefromcomplicatedbud-getaryaccountingpracticestononinclusionofcorporateandbankre-structuringexpenses,incurredduringfinancialcrises,intothebudget.NotmanyattemptshavebeenmadeintheliteraturetofindouthowTheauthorsaregratefultoHanaPolackovaBrixi,SaraCalvo,MichaelDooley,WilliamEasterly,IndermitGill,SantiagoHerrera,AshokaMody,SaumyaMitra,GuillermoPerry,JohnWilliamson,participantsoftheEcono-mists’Forum1999,andananonymousrefereeformanyinsightfulcomments.32HomiKharasandDeepakMishramuchdifferenceitwouldmaketothereportedbudgetdeficitifsomeoralltheseproblemsaretakenintoaccount.Asaconsequence,conven-tionalbudgetdeficitsarerarelyfoundtobeusefultoexplainmacroeco-nomicissues,betheytheinflationaryimpactofbudgetdeficitordiscussionoflong-termgrowth.Thedeficiencyoftheconventionalbudgetdeficittoaccuratelyre-flectthefiscalpositionofaneconomyismostevidentinthecurrency-crisisliterature.Oneofthelong-standingexplanationsofcurrencycri-sesisthattheyresultfromprofligatefiscalpolicies(Krugman1979).Anaturalimplicationofthishypothesisisthatthecollapseofafixedex-changerateisprecededbylargeandpersistentbudgetdeficits.1Eventhemostardentcriticsofthishypothesisagreethatmanycurrencycri-sesinthedevelopingcountriesduringthe1970sand1980s,andsomeoftherecentcrises,likeBrazil’sdevaluationin1999,havebeendrivenprimarilybyindefensiblefiscalpolicies.Surprisingly,however,thelargebodyofempiricalworkaimedatexplainingcurrencycriseshasfoundlittleevidenceofanysystematiclinkbetweencurrencycrisesandre-portedbudgetdeficits.2ThelackofevidencehasregrettablybeeninterpretedasarejectionofKrugman’smodel.We,however,arguethattheevidenceshouldbeinterpretedasyetanotherreasonwhy“budgetdeficit”asconvention-allymeasuredisinneedofafix.Thefollowingtwoobservationshaveledustomakesuchaclaim.First,weempiricallyconfirmtheimportanceoflookingbeyondthereportedbudgetdeficitinassessingthefiscalsituationinacountry,es-peciallyindevelopingcountries.Wedosobyprovidinganalternativedefinitionofbudgetdeficitthatisreferredtoastheactuarialbudgetdeficit.Unliketheconventionaldeficit,whichisaflowconcept,actu-arialdeficitiscomputedusingstockvariablesandisdefinedasthechangeinthestockofpublicdebtandthemoneybase(thatis,totalstockofgovernmentliabilities).Thequantitativedifferencebetweentheactuarialandtheconventionaldeficitsiscalledthehiddendeficit.Using1.Krugman’s(1979)modelhasbeenextendedinseveraldirectionstoshowthatcollapseofthepegmaybeprecededbydeteriorationofeconomicfunda-mentals,includingloosemonetarypolicy,bankfailures,andanappreciationoftherealexchangerate.Forarecentreviewoftheliterature,seeFloodandMarion(1997).2.Forexample,seeKaminsky,Lizondo,andReinhart(1997)andFrankelandRose(1996).FISCALPOLICY,HIDDENDEFICITS,ANDCURRENCYCRISES33datafrom32countries,from1980–97,weshowthattheaveragehiddendeficitsarerelativelylargeindevelopingcountriesascomparedtoin-dustrialcountries,andinsomecountriesarelargerthaneventheircon-ventionalbudgetdeficits.Second,wedemonstrateempiricallythatthereisacloselinkbetweenthenumberofcurrencycrisesandhiddenandactuarialdeficits.Thislinkisnonexistentifoneusesconventionaldefi-citsinplaceofactuarialdeficits.Theconceptofhiddendeficitsanditsassociationwithcurrencycri-seshasimportantpolicyimplications.First,anincreaseinhiddendefi-citsisfoundtoaffecttheexpectationsofmarketparticipantsinthesamewayasanincreaseinvisibledeficitsofthesameamount,soanyat-temptonthepartofthepolicymakertokeepthevisibledeficitssmallbyrunninglargehiddendeficitsisfutile.Second,whiledesigningpoli-ciestoreducethelikelihoodofcurrencycrises,oneshouldbearinmindthatcurrencycrisesmaybecausedbypasthiddendeficits,aswellasbycurrentandprospectivehiddendeficits.3Althoughitisdifficulttodistinguishthesetwocases,recentresearchshowsthatpolicymakers,usingnonperformingloansofthebankingsystemandmovementofbankshares,maybeabletodeducetheapproximatesizeofprospectivehiddendeficits.Third,hiddendeficits,whenaveragedoveralongpe-riod,say20–30years,provideacrudeapproximationofhowmuchcon-tingentclaimsandcapitalgainsandlossesthegovernmentislikelytoabsorbinthelongrun.Therefore,anefficientintertemporalbudgetaryallocationcallsformakingprovisioninthebudgeteveryyearforanamountequaltoitslong-runaveragehiddendeficit,sothatthecountrycanmeetfuturecontingentclaimsandcapitalgains