The Use of the Parallel Market Rate as a Guide to Setting the Official Exchange Rate

Boardof Governorsof theFederalReserveSystem InternationalFinanceDiscussionPapers Number564 September1996 THEUSEOFTHEPARALLELMARKETRATEAS A GUIDETO SEn~G THEOFFICIALEXCHANGERATE NitaGheiandStevenB.Kamin NOTE: InternationalFinanceDiscussionPapersarepreliminarymaterialscirculatedto stimulate discussionandcriticalcomment. Referencesinpublicationsto InternationalFinanceDiscussion Papers(otherthanan acknowledgmentthatthewriterhashadaccessto unpublishedmaterial)should beclearedwiththeauthoror authors.

ABSTRACT Thispaperaddressesthemeritsof usingtheparallelexchangerateasa guideto settingthe officialexchangerate. Ideally,policymakerswouldsettheexchangerateatthelevelthatwould balancetradeandsustainablecapitalflows--thatlevelisreferredto astheequilibriumexchangerate. In practice,itisdifficultto identi~ theequilibriumexchangerate,particularlyincountriesthathave experiencedmacroeconomicvolatilityand/orstructuralchange. In thiscontext,whereparallelmarkets for foreignexchangeexist,it isnaturalto considertheparallelrateasa proxyfortheequilibrium exchangerate,sinceit issetdirectlybythemarket. Thepaperdevelopsananalyticmodelto explore therelationshipbetweentheparallelexchangerateandtheequilibriumrate. It isdeterminedthatonly undera fairlynarrowsetof circumstanceswilltheparallelratebesetata levelcloseto the equilibriumexchangerate. Thepaperthencomparestheevolutionof officialandparallelexchange ratesovertime,ina largesampleof differentcountries,to providea feelfortheapplicabilityofthe previously-derivedtheoreticalresults.

TheUseoftheParallelMarketRateasaGuidetoSettingtheOfflcialExchangeRate NitaGheiandStevenB.Kamin* I.Introduction Determiningtheappropriatelevelatwhichtosettheexchangerateisachallengingproblemforany countrypursuingamanagedorfixedexchange-ratepolicy.Ideally,acountrywouldsetitsexchangerateatthe realequ.ilibnumra~t,hatis,therateconsistentwithinternalandexternalbalance(tie latterreferringtobalance betwmntradeandsustainableeapital-accounfltows).Eveninrelativelystableandmatureindustrialeconomies, howevert,herealequilibriumleveloftheexchangerateisusuallydifficulttoidenti~. Indevelopingcountries subjecttomacroeconomicinstabilityand/orstructuralchange,thisidentificationisevenmoredifficult.The determinatioonftheequilibriumrealexchangerateisespeciallyuncertainiftheeconomyisinthemidstoftrade liberalizatioanndotherreformsthatpromisetochangepreviouslyexistingrelationsbetweentradeperformance andtheexchangerate. Inacontextwhereaparallelmarketforforeignexchangeexists,itmayappearnaturaltoconsiderthe pardel exchangerateasaproxyfortheequilibriumrealexchangerate. Theparalielexchangerateusuallyhas thebenefitofbeingdeterminedinafreemarket,andhenceisnotobviouslycontaminatedbythedistortionary effecwofgovernmenptolicy.Mdtipleexchangeratemanagementsf,ormalandinformal,legalandillegal,were thenormfordevelopingcountri=untilveryrecently.Eventhoughanincreasingnumberofcountiieshave tiled theirexchangera@,oftenaspartofalargerliberalizationeffort,parallelforeignexchangemarketshave l~e au~orsMecomespnden~me EconomicTimes~d SeniorEconomist,Divisionof~~rnation~ Finance,BoardofGovernorsoftheFederalReserveSystemrespectivelyW. earegratefultoLawrence Hinkle,PeterMontielandSteveO’Comellforhelpti mmmentsandsuggestionsT. hispaperwasprepared asachapterinaneditedvolume,LawrenceHinkleandPeterMontiel,eds,ficllange RafeManagementin bw InconleCotlntnes:Metlzoblogiesfor EnzpiricalAnalysis,underpreparationbytheWorldBankThe viewsexpressedinthispaperareoursanddonotnecessarilyreflectthoseheldbytheBoardofGovernors, theWorldBank,TheEconomicTimesorothermembersoftheirstaffs.Weareresponsibleforanyandall emors.

Boardof GovernorsoftheFederalReserveSystem InternationalFinanceDiscussionPapers Number564 September1996 THEUSEOFTHEPARALLELMARKETRATEASA GUIDETO SETTINGTHEOFFICIALEXCHANGERATE NitaGheiandStevenB.Kamin NOTE: InternationalFinanceDiscussionPapersareprelimina~ materialscirculatedto stimulate discussionandcriticalcomment. Referencesinpublicationsto InternationalFinanceDiscussion Papers(otherthananacknowledgmentthatthewriterhashadaccessto unpublishedmaterial)should beclearedwiththeauthoror authors.

ABSTRACT Thispaperaddressesthemeritsof usingtheparallelexchangerateasa guideto settingthe officialexchangerate. Ideally,policymakerswouldsettheexchangerateatthe levelthatwould balancetradeandsustainablecapitalflows--thatlevelisreferredto astheequilibriumexchangerate. In practice,itisdifficultto identi~ theequilibriumexchangerate,particularlyincountriesthathave experiencedmacroeconomicvolatilityand/orstructuralchange. In thiscontext,whereparallelmarkets for foreignexchangeexist,it isnaturalto considertheparallelrateasa proxyfortheequilibrium exchangerate,sinceit issetdirectlybythemarket. Thepaperdevelopsananalyticmodelto explore therelationshipbetweentheparallelexchangerateandtheequilibriumrate. It isdeterminedthatonly undera fairlynarrowsetof circumstanceswilltheparallelratebesetata levelcloseto the equilibriumexchangerate. Thepaperthencomparestheevolutionofofficialandparallelexchange ratesovertime,in a largesampleof differentcountries,to providea feelfortheapplicabilityofthe previously-derivedtheoreticalresults.

TheUseoftheParallelMarketRateasaGuidetoSettingtheOfficialExchangeRate NitaGheiandStevenB.Kaminl I.Introduction Determiningtheappropriatelevelatwhichtosettheexchangerateisachallengingproblemforany countrypursuingamanagedorfixedexchange-ratepolicy.Ideally,acountrywodd setitsexchangerateatthe realequilibriumrate,thatis,therateconsistentwithinternalandexternalbalance(thelatterreferringtobalance betw=ntradeandsustainable~pit.al-accounftlows).Eveninrelativelystableandmatureindustrialeconomies, howevert,herealequilibriumleveloftheexchangerateisusuallydifficulttoidenti&.Indevelopingcountries subjecttomacroeconomicinstabilityand/orstructuralchange,thisidentificationisevenmoredifficuh.The determinatioonftheequilibriumrealexehangerateisespeciallyuncertainiftheemnomyisinthemidstoftrade liberalizatioanndotherreformsthatpromisetochangepreviouslyexistingrelationsbetweentradeperformance andtheexchangerate. Inacontextwhereaparallelmarketforforeignexchangeexists,itmayappearnaturaltoconsiderthe paralielexehangerateasaproxyfortheequilibriumrealexchangerate. Thepartilelexchangerateusuallyhas thebenefitofbeingdeterminedinafreemarket,andhenmisnotobviouslycontaminatedbythedistortionary effectsofgovernmenptolicy.Multipleexchangeratearrangementsf,ormalandinformal,legalandillegal,were thenormfordevelopingcountriesuntilveryrecently.Eventhoughanincreasingnumberofcountrieshave tiled theirexehangerab, oftenaspartofalargerliberalizationeffort,parallelforeignexchangemarketshave l~e auhors~e cofiespnden~ me EconomicTimesandSeniorEconotist, Divisionof~~mation~ Finance,BoardofGovernorsoftheFederalReserveSystemrespectivelyW. earegrateful10Lawrence Hinkle,PeterMontielandSteveO’Comellforhelp~ commentsandsuggestionsT. hispaperwasprepared asachapterinaneditedvolume,LawrenceHinkleandPeterMontiel,eds,&clzangeRafeManagenzenitn bw InconleColtntries:A4et)lodologiefsor EnlpiricalAnalysis,underpreparationbytheWorldBank.The viewsexpressedinthispaperareoursanddonotnecessarilyreflectthoseheldbytheBoardofGovernors, theWorldBank,TheEconomicTimesorothermembersoftheirstaffs.Weareresponsibleforanyandall enors.

2 notdisappearedas~ Nigeria,whichhasneverSUCwssfulluynifieditsexchangerate,isaprominentexample inAfrica;inVenezuela,whichunifieditsexchangeratein 1989,aparallelmarketemergedin1994following thereimpositionofcapitalcontrols. Notwithstandingtheappealofa market-determinedexchangerateas a guidetosettingtheofficial exchangerate,howeverv,ariousfactorscomplicatetherelationshipbetweentheparallelmarketrateandthereal equilibriumexchangerateinaneconomy.First,whiletheparallelmarketforforeignexchangemaynotitselfbe controlledbythegovemme4n~nditionsinthatmarketarelikelytobeaffated bygovernmentpolicy.Relative suppliesanddemandsforforeigncurrencyintie parallelmarketwillbealteredbytheleveloftheofficial exchangerate,theextenttowhichexchangeand&adecontrolsareenforced,andthegovernment’fsormulafor rationingforeignexchangereceiptstoimporters.Second,becausethep@lel exchangemarketrepresentsan assetmarketaswellasatrade-relatedmarke4theparallelmarketrateislikelytoreflectexpectationsp,olitical concernsc,apitalfigh~andotherfactorsnotdirectlyassociatedwiththerealequilibriumexchangerate. Hence, onlyunderarelativelynarrowsetofcircumstanceswilltheparallelmarketrateserveas a useti guideto determiningtie equilibriumvalueoftheofficialexchangerate. Inthispaper,weexplorethedeterminan~oftheparallelmarketrateinordertoass~s ifiusetiness as a guidetomeasuringtheequilibriumvalueof theofficial realexchangerate. Theplanofthispaperisas follows.InSectionII,parallelexchangemarketsaredefinedmorespecificallyandvariouscharacteristicof parallelexchangemartiaredestibed.2 SectionIIIaddrmsestheissueoftheunificationofexchangemarke@ andtheappropriatetargetfortheofficialexchangerate.InSectionIVwereviewasimpletheoreticalmodelof parallelexchangemarketstoshedlightonhowparallelmarketratesaredeterminedinrelationtobothofficial exchangeratesandequilibriumexchangerates. SectionVcomparestheevolutionofparaUelandofficialreal exchanger- overtimetoprovideaf~l fortheapplicabilityofthetheoreticalresultspresentedintheprevious .. ZForgener~smeys oftie issuesassocia~ withp~~el marketsforforeignexchangessee‘genor (1992)andKiguelandO’Comell(1995).

3 section.SectionVIconcludes. II. EssentialCharacteristicosfParallelExchangeMarkets 11.1BmicDefinitions Aparallelforeignexchangemarketsystemisoneinwhichtransactionstakeplaceatmorethanone exchangerateandatleastoneoftheprevailingratesisatieelyfloating,market-determinerdate(theparallel exchangerate).(KiguelandO’Connel1l,995)ParaUeml arketsystemsrepresentasubsetofthebroadercategory ofmultipleexchangerateregimes,whichrefertoanyregimesinwhichtwoormoreexchangeratesareapplied tothesa,mecurrencyM, anydevelopingmuntrieshaveappliedseparate,fixedexchangeratestodifferenttypes oftransactionsb,utthispracticeis,inessence,equivalenttoasingleexchangeratecoupledwithdifferenttaxa orsubsidi=(dependinognthetransaction).Bycontrast,aparallelmarketforforeignexchangeisdistinguished bythefactthattheparallelexchangerateisdeterminedfreelyinthemarke~Usually,theofficialexchangerate in parallelmarketsystemsispeggedbytheauthoritiesat a particularfixed(orcrawling)rate,althoughin principletheoficialratecouldbeflotig aswell,Additionallyi,tisfrequently-dthoughnotalways-thecase thattheoficialexchangerateapplimtocurrentaccounttransaction,whiletheparallelmarketrate,whetherlegal orillegal,appliestocapitaIaccounttransactions. Parallelmarketsfmforeignexchangecane~ge onlywhenthegovernmentimposesexchangecontrols, thatis,restrictionson thevolumeof certainforeignexchangetransactionsor on thepriceat whichsuch transactionsaremade, Tradebarriers,quantitativerestrictions,orhightariffsalonearenotinthemselves sufficienttogiverisetoaparallelexchangemarket.Whilesuchcontrolsmayaffectthedemandorsupplyof foreigncurrencies,theywillnotdriveawedgebetweenexchangeratesfordifferenttransactions,aslongas foreignexchangeisfreelyavailableforalltransactionsatanofficialormarket-determineedxchangerate. A parallelmarketarisawhenthegovernmenltimitstheamountofforeignexchangethatcanbeboughtorsold for particulartransactionsc,ausingexcessdemandorsupplytospilloverintoaparallelmarkegorauthorizesthat exchangeratesforcertaintransactionsbepeggedandforothertransactionsbefloating.

4 11.2kgal andIllegalSystem Parallelexchangeratesystemsmaybelegalorillegal.Whentheparallelmarketforforeignexchange is legal,it is oftenreferredtoas adualexchangerate(DER)system.Inthesecases,mostcurrentaccount transactionstakeplaceatapeggedcommercialrate,andcapitalaccount&ansact.ionastamarketdetermined financiarlate.AnumberofcountrieshaveexperimentedwithDERsystemsofvaryingduration.Somecountriw maintainedofficialdualexchangeratesforlongtimeperiods,suchasBelgium(from1957to 1990)andthe DominicanRepublic(until1993).Theparallelmarket,inthesewuntria, wasusedtoinsulatether=tofthe -nomy fromshorttermcapitalflows.France(1971-74)andItaly(1973-74)adopteddualratesforashort periodfollowingthemllapseoftheBrettonWoodssystemasatransitorymeasure.ArgentinaMexiwand VenezuelaadoptedDERregimes inthe1980sh thewakeof balanceofpaymentscrisesandhugecapital outflows. Illegalparallelmarketsystemsemergewhenprivateagenfiattempttoevaderestrictionsonthepriceor quantityofforeignexchangetransactionsI.llegalmarketswerethenorminmostofAfri~ andSouthAsia,as wellinseveralLatinAmericanmuntrim,~peciallythroughthe 1980s.Theparallelmarketsweregenerally toleratedbytheauthorities,withsomeexceptions.Forexample,thethreatofenforcementandpenaltieswas significanitnGhanapriorto1983,buttheseeffortsfellbythewaysidelateron,andthecoverageoftheparallel marketgrew,asdidtheparallelpremium(KiguealndO’Connell,1995).InSudan,tradingontheparallelmarket wasacapitaloffence,andenforcemenwt asattemptedbetween1970and1990.Buteventhethreatofcapital punishmentdidnottotallywipeouttheparallelmarket,thoughitmayhavekn afactorintheveryhigh p~miLlmobservedinSudan. Inprinciplet,hereislittledifference,intermsofmacroeconomicimplicationsb,etweenlegalandillegal systems. Ineithercase,free-markettransactionsinforeignexchangetakeplacealongsidecontrolledpricetransacdonsT. hismeanshat ineitherlegaloriliegalsystems,thereareincentivesfortransactionstospillover or “leak”hornonemarketintotheother. Theseleakagesmaytendtounderminetheparallelexchangerate

5 system,dependinguponhowrigidlyexchangecontrolsareenforced. 11.3:ParallelMarketsinthe1990s Agreatmanycountri=haveexperimeMn withparallelexchangearrangemen~atvarioustimes.These arrangementhsaveincludedformal,legaldualratesaswellasillegal“black”rata. However,theincidenmof sud wangernentshasbeendeclininginthe1990s,asanincreasingnumberofdevelopingcountrieshavesought tounifytheirexchangerati, oftenaspartofalargerstructuralreformeffortwhichincludesliberalizationofthe externalaccounts.Evenpartialconvertibilityofthecurrency,forment accountpurposes,forexample,leads toprogresstowardsunifyingtheexchangerate. Observersfrequentlyviewtheincidenceofrestrictionsoninternationaltransactionsasevidenceof prevalenmandimportancoefparalleelxtige marti. AccordingtoIMFrepo~, aboutone-haifofthemember countrimimposer~trictionsonpaymen~ontransactiononthecurrentaccount;overthree-quartersdosoon capitalaccountpayments(Table1).However,themereexistenceofrestrictionsdoesnotnecessarilyimplythe existenceofsi@lcantparallelmarb~, sincetheseareqtitative data,withtwovalues-yesandno,andsodo notcaptureeithertheintensityofrestrictionsortheeffectivenessofenforcementTherefore,usingtheexistence on paymentsrestrictionswodd r~ult inanoverestimateoftheprevalenceofparallelmarketsforeforeign exchange. Parallelmarketsarelikelyto beunirnportan~andtheparallelpremiumlow,whenthepayments r~trictionsandcapitalcon@olsareeitherminimalornotenford. Forexample,SouthMncaimposedcapital controlsin1985,followingmassivecapitaloutflows,andre-introdud adualexchangeratesystematthattime. Buttheparallelpremiumhasremainedmodest-themedianpremiumwas4.4percentfortheperiod1980-89, anddeclinedto2.3percentduringtheperiod1990-94(Ghei,KiguelandO’ComeU1996)-whichwouldbe consideredtoaunifiedexchangerateregimeunderthedeftition inSection5below.Thereareseveralother examplesof countri~withextremelylowpremia,including~ailand, MalaysiaandIndonesia,withmedian premiavaryingfmm-1.5to3.4percent(Gheietal,ibid).

6 Table1 IncidenceofPaymentsRestrictionsamongIMFMemberCountries 1980 1994 CurrentAccount 51.77% 51.69% CaDitalAccount 78.01% 77.53% SourceE:xchangReestrictioannsdExcbangAerrangcmenAUn:nuaRleporotftheIMFv,ariouisssues. Atthepr~entmomen~parallelexchangeratearrangementaretobefoundindevelopingcountriesonly; Belgium,whichwasthelastdevelopedcountrywithdualexchangerates,movedtoaunifid exchangeratein 1990.Parallelexchangerateswereafeatureofonehalfofour sampleoftwenty-fourdevelopingcountriesat theendof1994(seeTable2)3.Parallelexchangeratearrangemen&wereevenmorewidespreadindeveloping countriesin the1980s-everycountryinoursamplehadmorethanoneexchangeratein 1985. Oursample includescountrieswheresignificantparallelmarkefiexistedforsometime,andincludesmostof themore importantdevelopingcountries,outsideeasternand=ntral EuropeandtheformerSovietUnion. Parallelmarkefihavehaddifferentdegreesoflongevity.ArgentinaMexicoandVenezuelahadlegal dualratesthatwereexpectedtobetemporary.Allthreecreateddualratesandthenunifiedwithintheperiod, 1980to 1994,thoughaparallelmarketdidre-emergeinVenezuelain1994,asdiscussedabove.OtherLatin Americancountriesmovedto multipleratesor unifiedwithinthesameperiod.In the fican andAsian muntri=,bycontras~paraLlerlnar~ weremorelonglivd Afewofthesecountriesunifiedtheirexchangeratw inthe1990s,butmajorexceptionsremain-mostlyinMCZ includingNigeri~KenyaandZambia. Theleveloftheparallelpremiumhasdeueased,onaverage,inthecountrieswitiparallelexchangerate wangements.Wefindlowerprerniain1994reladveto1985inrnanycases,foroursample.Foraselectedgroup . me sampleisdrawntim aWorldBankresearchproject,“Macroeu)nomicImplicationsofMultiple ExchangeRatesinDevelopingCountriW”.Avolumebasedontheprojec~“ParalielExchangeRatesin DevelopingCountries,M.A.Kiguel,J.S.LizondoandS.A.O’Comell,eds,isforthcoming.Thedatawere updatedtotheendof1994forthispaper.

7 Table2 StatusofParallelMarket/LevelofParallelPremium(%) ? country 1980 1985 1990 lm htin AmericanandTurkey Argentina Unified 30.79 29.93 unified Bolivia 19.85 223.60 unified unified Brazil 8.90 30.12 14.28 Unifi Chile 6.03 25.39 16.78 8.67 Colombia unified 11.42 9.24 6.12 DominicanRep. 38.00 7.69 68.01 unified Ecuador 11.45 76.90 23.53 5.45 Mexico unified 28.46 7.41 Unifld Peru 36.25 29.53 104.80 unified Uru~ay unified 9.41 10.93 16.51 Venezuela unified 104.00 unified 4.73 Turkey 9.98 -7.17 unified unified AflicaandAsia gena 193.25 375.23 248.81 253.95 Egypt 7.77 122.90 89.47 unified Ethiopia 35.02 127.66 192.75 113.2 Ghana 485.92 143.48 9.84 Utifi Kenya 9.38 4.51 3.11 19.78 Malawi 92.27 49.51 17.51 14.62 Nigeria 67.67 306.22 19.40 231.87 Sudan 92.40 27.14 955.45 53.13 Tanzania 174.00 271.89 56.36 Unified Zambia 70.27 65.39 279.56 -6.15 India 9.58 15.12 9.10 Unifl Pakistan 26.26 -0.67 8.72 Unifi 1 . NOM:Premiumisdefinedas(&/e”-l)I*W.Exchangwe kexpressedaslocalcurrency/US$. Dataareannualaveragescalculatedusing endingofquartervalum. Source:IFSandWorldCurrencyYearbookC, urrencyAnalysis,variousissues

8 ofhighpremiumcountri=4,Ghei,KiguelandO’Comell(1996)fmdthatthemedianpremiumfortheperiod 1990to1994was49perm~ comparedwithafigureof100percentfortheperiod1980-89.Similartrendshave beenobsexvedformoderateandlowpremiummuntri= aswell. Ovmti,thereareindicationsthatdevelopingcountriesue movinginthedirectionofunifiedexchange rates.Thenumberofcountrimwithsignificantparallelmmke~hasdeclined,and thegapbetwun theofficial andtheparallelrateissteadilydecreasinginmostofthecountriesthatstillhaveparallelratm. 11.4RationaleforParallelMarht Systems Parallelmarketsystemsemergefordifferentreasonsindifferentcountri=. Thereisonelegitimate rationaleforasysteminwhichcment accounttransactionsareconductedatapeggedrateandcapitalaccount transactionasreconductedatafloatingrate:toinsulatedo-tic pricesandeconomicactivityfromexchangerate fluctuationsderivingtim transitoryshocksinthefinancialmarket. In practice,theimplementationofparallelmarketsystemsindevelopingmuntrimrarelyhasbeen consistentwiththisrationale.IncertainLatinAmericancountries,dualexchangeratesystemswereindeed adoptedinrmponsetostrong,temporarycapitaloutiowsr=uitingfrombtiance-of-paymentcsrisesinthe1980s, andthisdi&toa@ ex~ protecttheireconomiesfiornexcessive,transitorydepreciationsoftheexchange rate.However,thedualmarketswereretainedlongafterthefmancidcriseshadpassed.Moreover,evenafter thecriseshadpassed,thepdel ratescontinuedtobemoredepreciatedthantheo~cial rates;inadualrate syskmtiigned topm~ thewnomy tim exchangeratevariability-asopposedtoasystemdesignedtotarget theofficiarlateatalevelpe~iskntlymoreappreciakdthanwhatthemarketwodd set--theparallelratewould beexpectedtofluctuatebothaboveandbelowtheofficialrate. InMean countri~,p~el marketswereevenlessconsistentwiththebestradonalefortheexistence 4A~@ pre~um coun~ iswherethemedianp~tium eX=dS 50 per ~n~ The te~ “m~era~ premium”isappliedtocountrieswithmedianpremiumbetween10and50percen~Amedianlevelofless than10percentputsamuntryintothelowpremiumcategory.Thetimeperiodexaminedin1970-1994.

9 ofsuchs~terns.Frequentlye,xchangecontrolsweretightenedinth=e countriesasagradualovervaluationof theofficialexchangerateledtoex~ss demandforforeignexchangeattie officialrate. This,inturn,ledto creationofparallelmarketstoevadesuchcontrols,evenintheabsenceofstrongcapitalaccountpressures. Hence,exchangecontrolswereusedtopropuppersistentlymisalignedexchangerates,nottoinsula~the domesticeconomyfromtransitoryfluctuations. 11.5EconomicFunctionofParallelMarkets Wehaveidentiledtwocentraleconomicfunctionsofparallelexchangemarkets.Whenparallelrates emergedfollowingabalanceofpaymentscrisis,aswasthecaseinmuchofLatinAmerica,theparallelmarket wasusedprimarilytofinancecapitalflowsinandoutofthecountry.Therewasverylittlerationinginthemarket for tradetransactions,asforeignexchangesupplywasusuallyenoughtosatisfydemand. Onaverage,the premiumoftheparallelraterelativetotheofficialratewasquitemoderateinthesecases-thoughtherewere occasionaslpilcews henthepremiumwasveryhigh.Butthesespikesreflectedtemporarymacroeconomiccnsm, notadrasticandpersistentmisalignmenotftherealexchangerate. PreciselytheoppositewastrueinthecaseofmostAfricancountrim.Intheprototypicaclase,foreign exchangerationinggrewmorestringenotvertimeastheofficialexchangeratebecameincreasinglyovervalued. Importerswholackeda-s toeverscarcerforeignexchangethroughtheo~cial channels,turnedtotheparallel markettoobtainforeignexchangefortradetransactionsT, heparallelpremiumgrewtovery highlevels,and stayedthere,astheofficiarlate* moreandmoreovervaluedG. hanaisthekxtbookexampleforthis,when by theendof the 1980s,tie o~lcialexchangeratewasso overvaluedthatit karne irrelevantformost transactionse;vendomesticpricesandMation reflectedtheparallel,nottheofficialrate(ChibberandShtilk, 1991). III.UnificationofExchangeRatestoaSingleEquilibriumRate Observa haveidentiledvariousnegativeconsequencesofexchangecontrolsandtheparallelmarkets that theyengender. A non-exhaustivelist wodd include,fns~ thefactthatexchangecontrolsallowthe —--

10 authoritiatomaintainapersistentlymisalignedofficialexchanger~-perhaps coupledwithinappropriatefiscal, monetary,andcommercialpolicies--withoultosingalltheirinternationalreserves,therebydistortingrelative pricesin thewonomyandinhibitingthegrowthofexports. Second,Wause parallelmarketregimesof~n involvetherationingofforeignexchangeatsubsidizedratestothosewithpreferentialaccesstotheauthoriti~, exchangecontrolsencouragethedevelopmenotfrent-seekingbehavioramongpriva~entrepreneurs.Finally, theintroductioonfexchangecontrols,whichbytheirnaturearehardtoenforceandprofitabletoevade,tendsto promoteacultureoflawevasionamongprivateentrepreneursthatmayspilloverintootherareassuchastax complianceoradherencetootherwonomicandfiancial regulations. Inrmponsetotheseandoth~adverseeff~ ofexchangecontrols,anincreasingfractionofdeveloping countrimhavemovedtodismantleexchangecontrolsandunifytheirexchangemarkets.Someparallelmarkets wmeabandonedeitherbecausetheywerenolongerneeded(whenthecrisiswasover)orbecausetheywereno longereffmtivet,hatis,rampantevasionofextiangecontrolsunaed thesystem.Otherdevelopingcountries (includingTanzania,GhanaandIndia)movedtolegalizetheirparallelmarketsasatransitionalmeasurewhile easingrestrictionosncment accounttransactions-asasteponthepathtoMlcation oftheexchangerate. In thosecases,unificationhasbeenpartofalargerstructuralreformeffortaimedatliberalizingmarketsoverall. Becausetheemergaceofaparallelmarketusuallyrefl~ theexistenceofanexcessdemandforforeign exchangeattheprevailingofficialexchangerate(sothatforeignexchangemustberationedbytheauthorities), successfulunificationoftheexchmgemarketsgenerallyrequiresthattheofficialexchangeratebedevalued stilcientlytoe~” “ excessdemandsforforeignexchange.Putanotherway,successfulWlcation requirw thattherealofficialexchangeratebesetatits“realequilibrium”level. Definitionsof the“realequilibriumexchangerate”aboundintheliterature.Herewefocusontwo equilibriumconceptst,he“short-runa”nd“long-run”equilibriumexchangerates,whicharemorefullyd~cribed inanotherchapterinthisvolumes Thelong-runequilibriumexchmgerateisthatratewhichequatesthetrade . bpekrMontiel,“me maw oftie ~ng-R~ EquilibriumRealExch~ge‘a~”*

11 balancewithsustainablenetcapiti flOWS,when fisc~, mOne@, mcl commerciaplolici=are“normal”and sustainablew,henforeignassetholdingsareequaltotheirdesiredlevels,andwhennoexchangecontrolsarein place.Theshort-runequilibriumexchangerate,bycontras~istheleveloftheexchangeratethatequatesthe tradebalancetonetcapitalflows--thatis,equatesthesupplyanddemandforforeignexchange--givencwent settingsofeconomicpolicyandcment demandsforforeigncapiti, regardlessofwhethereitherareattheir normallevels;asinthecaseofthelong-runequilibriumrate,theshort-runequilibriumrealexchangerateis definedasthatwhichinducesbalance-of-paymenetsquilibriumwhennoexchangecontrolsareinplace. Theshort-runquilibriumrd exchangeratemaydiffertim itslong-runvalueforinnumerablereasons. Forexamples,tartingoutinaneconomywhereboththeexchangera~ andothergovernmentpoliciesareattheir sustainablel,ong-runsettingst,heimpositionofextraordinaryimportbarriers,byreducingtheeffwtivedemand fmforeignexchangeintheforeignexchangemarke~willcausetheshort-runequilibriumrealexchangerateto appreciaterelativetoitsiong-runlevel.Converselye,xpectationsofa futuredevaluationorpoliticaldisruption willleadto a temporaryheighteneddemandforforeignassetsthatwillleadtheshort-runequilibriumreal exchangeratetodepreciateinrelationtothelong-runequilibriumra~. Bydeftition,aslongastheoff~ciaelxchangerateisatitsshort-runequilibriumlevel,therewillbeno imbalancebetweendemandsandsuppliesforforeignexchange.Thereforeparallelexchangemarkets,totie extentthattheyreflectanexcessdemandforforeignexchangeunderuent economicpolicim,areassociated withrealofficialexchangeratesthatareovervalued--thaits,apprwiated--withrespecttotheshort-runreal equilibriumexchangerate. However,successfullysustainingaunificationoftheexchangemarketprobably requirmthattherealofficialratebesettoitslong-runequilibriumvalue,notitsshon-runvalue.Moreover,it probablyrequiresthatotherelementsofgovernmenptolicy,inadditiontotheexchangerate,bechangedaswell. Toseethis,considearnexampleinwhichtheauthoritiehsaveimposedveryhighquantitativerestrictions (QRs)onimportst,herebyreducingtheeffectivedemandfmforeignexchangeandappreciatingtheshofi-runreal equilibriumexchangeraterelativetoitslong-runlevel.Atthesametime,assumethattheauthoritieshaveaiso

12 w theofficiaelx-ge rateatalevelthatisevenmoreappreciatedthantheshort-runequilibriumrate,sothat in responsetoanexcessdemandforforeignexchange,theauthoritiesrationdollarsanda parallelmarket ernerg=.Inthisexampled,evaluingtheofficialexchangeratetoitsshort-runequilibriumlevelwouiduni~ the extiangemarketsb,utwouldleavemuchofthedistortionassociatedwiththeQRsuntouched.A~lly successful tilcation thereforewouldentaildismantlingtheQRsanddevaluingthereaIofficialexchangeratealltheway toitslong-runequilibriumlevel. k practices,omecountri=mightnotregarditasfeasible,withinashortperiodoftime,bothtodevalue theirexchangeratetoitslong-runequilibriumlevelandtoadjustinappropriatefiscal,monetary,andcommercial policiesas well. Theymay fearthe inflationaryad distribution consequencesof a verylarge“maxi” devaluationa,ndmayfacestrongpoliticalpressuresagainstchangingparticularaspectsofgovernmentpolicy, particularlytradebtiers thatbenefitcertainvated interests.Insuchacase,theauthoritiesmightatmmptto implemenatgradualprocessofunificadon,movingtheofficialratetowarditslong-runequilibriumvalueeven astheyadjustassociatedeconomicpoliciesintheappropriatedirectionaswell. Whethertheauthoritiesintendtoimplementeitheragradualorarapidtilcation, theywillneedto iden~ thelong-runequilibriumvalueoftherealexchangerate,and,asnotedintheintroductiont,histaskwill be fraughtwithmnsiderableuncertainty. Underthesecircumstancest,heprevailingparallelexchangerate W-M anobvio~PmXYfmthe~uilibriumrate,andtheauthoritiesmightnaturallyconsidertheparallelrate tobem appropria~targettowardwhichtomovetheofficialrate,eithergraduallyorallatonce.Asshownin thefollowingsection,however,therearemanyfactorsthatcodd causetheparallelra~ todivergesignticantly hornitslong-runequilibriumvalue,makingiginmanyinstan~, aninappropriatetargetfortheofficialexchange rate. IV.ASimpleModelofParallelExchangeRateDetermination Thisstion presentsasimplepartialequilibriummodeltoillustratehowtheparallelwket exchange rateisdeterminedinrelationbothtotheoKlciaelxchangreateandtothelong-runequilibriumexchangerate,that

13 is, therate thatwouldproduceequilibriumin thebalanceof paymen~undernormal,sustainablepolicy conditions!Thereisnoconsensursegardingthemostappropriatemodeltouseinexplainingtie parallelmarket rate,just as thereisnoagreementastowhichmodelbestexplainsthemovementoffloatingexchangerates amongindustrialcountries.Themodelillustratedbelowhastheadvantageofbeingrelativelystraightforward andintuitive,whilehopefillyhighlightingthemostimportantfeatur=influencingtheparallelmarketrate. N.1 BasicSetup Considerasmallopeneconomytradingintwogoods,anondomesticallyconsumedexportgoodanda non-domestictiyproducedimport;theworldpricesofbothgoodsarefixedandsettounity. Tofocuson developmentsintheexternalsector,weassumeittobesmallrelativetothedomesticeconomy,sothatthe analysis&cribestheoperationoftheparallelexchangemarketinpartialequilibriumTherefore,theoutputof anon-tradedgoodanditipriceareconsideredf~ed aswell.WeassumeforconveniencethattheUSdollaris theonlyforeigncurrencytraded. Turningtotheparametersofgovernmenptolicy,itisassumedthatmonetaryandfiscalpoliciwareat theirlong-~ sustainablleevels,andmoreovmf,ortiytid convenienceth,erearenotariffs,subsidies,orother commerciaplolicyintervention.(Theroleofimportbarrierswillbeexaminedlater.)Therefore,intermsofthe deftitionsintrodud intheprecedingsectiontheshort-runrealequilibriumvalueoftheexchangerateis,atleast initiallyt,hesameasitslong-runvalue.TheofficialexchangerateE(measuredintermsofdomesticcwency per dollar)ispeggedatanovervaluedlevelrelativetotheequilibriumrate. Therefore,atthatlevelofthe exchangerate,theflowdemandfordollars(tobeelaboratedbelow)exc~ theirflowsupply.Exportersare requiredtosurrendertheirdollarearningstothecentralbankattheofficialrateE. Importerspurchasedollars fromthecentralb~ whichrationsdollarsalesOS,basedontheamountofexportrevenuessurrenderedtothe centralbankOX,accordingtothecentralbankrationingfunction: %e modelanditsexpositionarebasedontheanalysispresentedinKamin(1993,1995).

14 0s = os(o~,osf()>0 (1) Inraponsetotheprevailingexms demandfordollarsattheofficialrateE,aparallelmarketfordollars pricedattheparallelmartirate Epemerges.Wefollowtheconventionasltock/flowapproachtoexchangerate determinationinpositingthatinthelongrun,theparallelratemovessoas@equateflowdemandsfordollars byimporterswithflowsuppliesfordollarsbyexporters.Thatis,ti thelongrunEpissetsoastobalancetie privatesector’scurrentaccount Intheshortrun,ontheotherhand,theparallelmarketrateisassumedtomove exclusiveltyosettheportfoliodemandfordollarsequaltothestockofdollarsoufitanding,sothatatanygiven momen~theprivatecwnt accountmaybeoutofbalance. IV.2me ParallelMarketRateinhng RunCurrentAccountEquilibrium Wenowanaiyzethedete* “onoftheparallelmartirate inprivatecurrentaccountequilibrium.The privatesectorcurrentaccountisdefied asthedifferencebetweenprivatedollarinflowsorsuppliesS and outflowsordemandsD. Weassumethatforeignersholdnodomesticassets,sothatchangesinthestockof dollarsheldbytheprivatesector,B,- exclusivelythroughimbalancesintheprivatesector’scurrentaccount: dB =S - D (2) Thecment account(orflow)demandfordollarsisaderiveddemandforimportedgoods.Arbitrage ensuresthatthepriceoftheimportwillbethesame,whetherpurchasedhornalegalimporterwithaccmsto officialforeignexchangeorhornasmugglerusingdollarspurchasedintheparallelmarket.’Ineithercase,the priceofimportswillbesetqual toi~marginaml s~theparallelmarketrateEP(sin@byassumptiont,ie foreign currencypriceissettounity).Therefore,theprivatedemandforimports,asindica~ inequation3below, TWe~sme hat theremen. t~ffs andthatrestrictionsonsmugglhg~ cosfl~slYevad~” ‘e ‘at&r assumptionisrelaxedbelow.

15 depend(nsegativelyu)ponthedom=ticcmency primofimportsEPrelativetothepriceofnon-tradeablmPm.8 D = D(Ep/Pn) =D(ep), D/( ) <0 (3) where&is therealparallelexchangerate. Thement account(orflow)supplyofdollarsderiva bothfromunder-invoiceddollarearnings--that is,exportreceiptsnotturnedovertothecentralbank--andfromofficialdollarsalestoimportersOS. Notethat whileholdersofimportii~~ haveanincentivetoover-invoicet,hisdoesnotaddtothetotalsupplyofdollars totheprivatesector,whichisfixedbythecentralbank’srationingtiction (equation1). LetXrepresentthe quantityoftotalexportsandtotaldo~arrevenueasswell(sincetheworldpriceissettounity),while$ represents theshareoftotalexportproceedsdivertedtotheparallelmarke~Then s = ox + Os(oxi = $x + 0s((1 -(#)x) (4) Exporters~ domesticcurTencpyrofi~subjecttorisingmarginalcostsofproduction--whicwhe assumetoberelatedtothepriceofnon-tradedgoods--aswellasrisingcostsassociatedwiththeunder-invoicing share~.Wecanderivethesupplycurvefortotalexportsasafunctionoftheweightedaverageofthereal(nontradeablespricedeflated)official(e)andparallelmarketexchangerates(eP):g X =X(@eP + (1-@)e), X’( ) >0 (5) Theunderinvoicingshare$canbeshowntopositivelydependupontherealparallelmarketpremium: (6) 8~ imwfi demmd isa ~ction ofin~me aswell. Sin~, intis p~~ ~~ibrium ‘Odel$ p~ciple, “ incomeisconsideredtobefixed,wedonotincludeitexplicitlyinthedemandtiction. (1993)for‘ews’ 9SW K-

16 For a givenvaiueoftherealoficid ratee, auniquerealpdlel marketrate& willequatedollar demandsandsuppliminequilibrium: D(e~ =@X+OS(O~ (7) Thisequationw beMer simplifiedifwepositth~ onaverageandoveralongenoughtimeperiod,thecentral bankwillhaveroughlystablienternationarleserves.Therefore,wecanassumethatoveralongtimeperiod,the centralbankwillresellallsurrenderedexportreceiptsOX= (1+)Xtolicensedimporters,afteritextrac~any foreignexchangeneedsofthegovernment(assumedtobeinvarianttotheexchangerate)D~10 OS(O~ =OX - Dg =(1-$)X - Dg (8) Therefore,equadon7canberewrittenas D(ep) =+X +(1-+)X - Dg =X - D$ =x((#)ep+(l -@)e) - D8 (9) Figure1belowdepictsvariousdifferentequilibriaintheparallelexchangemarke~dependinguponthe valueofesetbytheauthoriti~.TheDDcurvedepictsequation3,theprivatedemandforforeignexchangeas atiction oftherealparallelrate&. TheSScurvedepictsthesupplyofforeignexchangetothemarketasa titionof ~; itslocationalsoisatitionof e,sinmboth&andeaffectthetotalquantityofexportssupplied. e*isthelong-mequilibriumrealexchangerate,thatis,theleveloftherealexchangeratethatwoddcl~ the market-settoti de~ds forforeignexchangeequaltototalsupplies-in aunifiedforeignexchangemarket. Notethatwhatheofficialexchangerateissetequaltoe*,theparallelratealsomustequale*.11Inotherwords, u llwe bow thiS‘w D(e*)= X(e*). Dg inaunifiedexchangernarke~andifinaparallelmarketsystem, D(ep)= X(W +(1-@~*)-Dg,

16a ep D epl ... .... .... + ......... ..................... e* s D,S Figure 1

17 whentheofficialexchangerateis set at its equilibriumvalue,thereis no current-accountmotivefor the emergenceofaparallelforeignexchangemarket,sincethereisnoexcessdemandforforeignexchangeattie officialrate. Wenowcomidertheeffecfiontherealparallelrateofarealappreciationoftheofficialrate. Assume thattheauthoritiesallowtheofficialrateetoappreciatetoanovervaluedlevelel<e*. Becausethislowersthe profiti-ilityofexports,thesupplycurveSSshiftsinwards,creatinganexmssdemandforforeignexchangeat thatrate.Thisputsupwardpressureontheforeignexchangevalueofthedollarinthe(nowemergent)par~el marke~causingthep@el exchangeratetodepreciatefrome*to &l.12 Hence,incaseswheretheemergenceoftheparallelmarketreflectstheovervaiuationoftheofflcia.1 commerciaelxchangerate,theparallel-ket rate,onaverage,islikelynotonlytobemoredepreciatedthanthe commercialrate,butit probablyis moredepreciatedthanthelong-runequilibriumexchangerate~ well. Various factors arelikelyto determinetheextentto whichtheparallelrate is moredepreciatedthatthe equilibriumexchangerate. AsFigure1makesobvious,themoreovervaluedtheofficialexchangerate,andso thegreatertheextenttowhichtheSScurveshiftsinwards,thegreaterwillbethegapbetweentheparallelrate andtheequilibriumrate. Italsoisstraightforwardtoshowthatthegapwillbegreater,themoreelasticare exportsandthelesselasticareimportswithrespecttotheexchangerate. Theex~t towhichexportsurxenderrequiremts areenforcedplaysakeyroleindetermini.gnthevalue oftheparallelexchangerate,aswell.Recallthattotalexportsarea functionofaweightedaverageofthereal officialandparalleelxchangerates.Ifforeignexchangeregulationsaretightlyenforced,exportunderinvoicing willbelimited,reducingtheunderinvoicingratio$ andtherebyincreasingtheweightplacedontheofficial exchangerate.Inthiscase,theovervaluationoftheofficialexchangeratedepressestotalexportssigtilcantly, reducingthesupplyofforeignexchangetotheparallelrnar~ anddepreciatingtherealparallelmarketexchange thenitisobviousbyinspectionthatthesecondequationissatisfiedforep=e*. l~s resdt is~mis~nt withtilatfoundbyNOW*(1984)”

18 ratesubstantiallyrelativetotheequilibriumrate. Converselyi.,fforeignexchangere@ationsarepoorlyenforcedandwidelyevaded,theunderinvoicing ratiowillbehigher,theweightedaverageexchangeratewillbemorefavorableforexporters,andtoti exports willnotbeasseverelydepressed.~s willleadtolesspressureontheparallelmarketexchangerateanda smallergaprelativetotheequilibriumrate.Atanextremea,stosomeextentoccurredinsomeWean countries, theo~cialexchangeratebecomessowidelyevadedastobecomeirrelevanttomosteconomicdecisions.Inthis context,mosttradeisroutedthroughtheundergroundeconomy,andtie parallelexchangeratemaybecomea reasonablyaccurateguidetothelong-runequilibriumra~. Finally,thevalueoftheparallelexchangerateinlong-runequilibriumislikelytobeinfluencedbythe ex~nttowhichbarrierstoimports(aboveandbeyondmerelyrationingofficialsalesofforeignexchangethrough exchangecontrols)areenforced.Theaboveanalysisassumesthatonceimportersacquireforeignexchange, Whethmofflcialolyrfiomtheblackmarke~theymayusethatfinancingtofreelyimportgoods.However,ifthe authoritiesimposehighimportbarriersandtheyarewell-enforceds,othatsmugglingiscostly,thiswillactto redumthedemandforforeignexchangeintheblackmarket--thaits,theDDcurveshowninFigure1woddshift inwardsandtothelek Thisreduceddemand,inturn,wouldcausetheshort-runrealequilibriumexchangerate to appreciaterelativetoitslong-runvalue,sincethelong-runequilibriumvalueoftherealexchangerateis predation theabsenceofextraordinarycommercialpolicimofthissofi Theimpositionofthebarriersalso wouldleadtherealparalleelxchangeratetoappreciatreelativetothecasewhereimportbarriersarenotenforced. Forstilcientlytightcontrolsonimports,therealparallelratecouldalsobemoreappreciatedthanthelong-run equilibriumrealrate. IV.3TheParallelMarketRateinShort-RunPortfolwEquilibrium Theresultsdescribedabovearelikely,atbes~toholdonaverageoverrelativelylongperiodsoftime. Intheveryshort~ thestockofdollarsheldbytie privatesectorisconsideredtobefixed,sinceitti time to accumulatoerdecumulatdeollarsthroughaent accounitmbalan~;duringthisshortrun,thepardel market

19 rateconventionalliysmodeledasbehgde~tied bytieportfolio-bueddemandfordollars.Thisportfolio demanddependsupontherelativeexpectedratesofreturntoholdingdollarsanddomesticassets,whichare Muend byanticipatedinflation,otheraspectsofmaaoeconornicperformance,andpoliticalevensaswell. Thevolatilityofsuchexzons largelyexplainsthehighvolatilityexhibitedbymostfreelyfloatingexchange rates,includingparallelmarketexchangerat=. Forasimpletheoreticaelxposition,assumethatprivatesectoragentsholdtwoasse~intheirportfolio, dollarsanddomesticcurrency.FollowingDombuschet.al.(1983),thedesiredratioofthedomesticcurrency valueofprivatesectordollarholdingstothenominaldomesticmoneysupply(M)ismodeledasatiction ofthe expectedrateofdepreciationoftheblackmarketrate(theAdeno~ percentagechange): or,dividingtiough bynon-tradedgoodsprices, (11) Thenotationfortherateofdepreciation;,P, ornitsanexpectationaltermtoreflecttheassumptionofperfect foresight. Theportfoliodemandfordollars(whentheparallelmarketrateisstable)tracesoutadownward slopingcurvein(4, B)spawasshowninFigure2. GiventhatBisconsideredfixedatanyonemomen~the levelofBdeterminesthelevelof&atthatmomen~ Inthelongrun,theparallelmarketrateandtheprivatestockofdollarholdingsaredeterminedbythe require~ts ofbothportfolioandcurrentaccounetquilibrium.Inadditiontotheportfolioequilibriumcondition hdbed above,Figure2depictsthelocusofpoinfiforwhichtheprivatecurrentaccountisinequilibriums,o thatthestockofdollarsBheldbytheprivatesectorisunchanging.This curve,denoteddBsO,isvertical,since

19a B dB=o o ep Figure 2

20 for anygivenofflcid exchangerate,a singlevalueof theparallelmarketrateepclearstheprivatecment account.1T3hepointwherethetwocurv=cross--thesteady-stateequilibrium--istheonlypointwhereboththe currentaccountisinequilibriumandthestockofdollarsheldbytheprivatesectorequalsitsportfoliodemand. Finallyt,heverticalandhorizontaml owsrepraentthedirtion ofmovementofBandepoukideofequilibrium, whilethediagonalline-the “stablesaddlepath’-’-indicatestie path by whichB and# convergetoward equilibrium,shodd theystartoutou~ideofequilibrium. Wenowconsidertheeffectsontheparallelmmketrateofariseininflation--sayf,romOto20percent-leadingtohigherratesofnorninaldepreciationoftheofficialandparallelexchangerates. Thisisanimpo~t example,sincemanycountri~thatimposedexchangecontrolsexperiencedincreasesininflationandother measuresofmacroeconomicvolatilityataboutthesametime,particularlyinLatinAmerica(andTurkey). AsshowninFigure3,theincreasedexpectedlevelof~ation-and henceofnominaldepreciationof theparallelmarketrate-leadsagentstodesiretoholdahigherratioofdollarstodom~ticcurrency,causingthe Portfoliobalanmcurvetoshifiupwards.Equilibriumdollarholdingssh.ifttim BOtoBl,whiletheequilibrium realparallelmarketrateremainsunchanged.However,inordertoaccumulateadditionaldollars,theprivate currentaccountmustshiftintosurplustemporarilyw, hichinturnrequiresatemporarydepreciationofthereal parallelexchangerate.Hence,atthemomentofincreasedinflationexpectationst,heparallelmarketratejumps tiominitialquilibriumat(l) tothenewstablesaddlepathat(2). Afterthis,theaccumulationofdollm through thecurrentaccountsurplusreversesthedepreciationoftherealparallelmarketrateuntilthesystemreturnsto equilibriumat(3). h theexampledepictedabove,thed paralleelxchangeratebecomes,foratime,moredepreciatedthan i~ownequilibriumlevel(conditionaolnthevalueoftherealofficialrate).(Asaresultofthesetemporarycapiti ltForsimplicity,~s ~~ysis abs~ac~horntheinterestpaymentsassociatedwi~ netas~t holdings?U wellashornthewealtheffectsofassetholdingsonimportdemand.Inthepresenceofinterat paymentsor wealtheffec~,thedB=Ocurvewouldnotbevertical,sincethevalueoftheparallelmarketexchangeratethat clearedtheprivawcurrentaccountwouIddependuponthestockofdollarholdingsB.

20a B dB=o BI ........................ ~p=zo BO........................ Ep=o ep Figure3

21 outflowst,heshort-runequilibriumexchangerateinaunifiedmarketalsowodddepreciaterelativetoitslong-run equilibriumlevel.)Sincetie equilibriumleveloftheparallelexchangerateislikely,asshowninSection111.2, to bemoredepreciatedthanthelong-runequilibriumrateinaunifiedexchangemarket,thismeansthatthe parallelrateduringperiodsofmacroeconomicvolatilityandcapitaloutflowswillbemoredepreciatedsti~. Hence,duringperiodsofmacroeconomicvolatilityandheavycapitaloutflows,theparallelrateislikelytobea particularlybiasedguidetosettingtheappropriateleveloftheofficialexchangerate. Infa~ becauseoftheassetmarketfunctionoftheparallelexchangemarket,theparallelmarketratecan tradeatalargepremiumovertheofficialrate,evenwhentheofficialrateisclosetoitslong-runequilibrium value,thatis,thevaluethatequilibra~thebalanceofpaymentsin“normal”macroeconomiccircumstances.As notedabove,his isbecauseduringperiodsofheavycapitaloutflows,theshort-runequilibriumrealexchange rate in aunifiedexchangemarketmaydepreciaterelativetoi~ long-runlevel;iftheofficialexchangerate remainsatthat(i.e.equilibrium)long-runlevel,anexcessdemandforforeiWexchangewilldevelopthatwill causeaparallelmarketpremiumtoemerge.TheLatinAmericancountriese’xperienceswithexchangecontrols mayfitthisscenario.Aswillbediscussedfurtherbelow,invariousofth=e countries,acombinationoffactors ledtoabalance-of-paymenctrisisinthe1980s.Thegovernmentsrespondedtothiscrisisbydepreciatingthe officiaelxchangerate,butbecauseofthesizeofthecapi~ outflowstriggeredbythecrism,theparallelmarket ratesinthesecountriesdepreciatedstillfurther. Finallyw, eshouldunderscorethef= thateveniftherealparallelmarketexchangerate,onaverageover longperiodsoftime,wereagoodindicatorofthelong-runequilibriumred exchangerate,thevalueofthereal parallelmarketrateatanysinglepointinh wodd likelybeanextremelyunreliableproxyforthatequilibrium rate.Thisisbecausetheparallelmarketexchangerate,likeanyotherassetprice,dependsuponhighlyvolatile @otio&-&, adham&i&eMti@y volatile.Thisvolatilitymaybeseenquiteeasilyinthechartsofthe officialandparallelexchanger~ presentedattheendofthispaper.Hen&,asidehornthefactthattheparallel rateislikelytobeabiasedindicatorofthelong-runequilibriumexchangerate,itisalso,unlessaveragedover

22 verylongperiods,likelytobeahighlyuncertainorinaccurateindicatoroftheequilibriumrateaswe11.14 V.TrendsinOficialandParallelRealExchangeRates Thetheoreticalmodeldescribedabovesuggeststhattheparallelexchangerateislikelytobemore depreciatetdhanthelong-runequilibriumrealexchangerate,unless(1)macroeconomicfactorsinducingcapital flightarenotpresen~(2)exchangecontrolsarepoorlyenformd,and/or(3)therearehighimportbarriersthat arewellenforced.Toevaluatethesehypothesesw, ewould,ideally,comparethepathoftheparallelexchange rateinvariouscountri~tothatofthelong-runequilibriumrealexchangerateinordertogaugetheextentto wtichtheparaIlerlatemayserveasause~ guidetodetermininggtheequilibriumexchangerateand,therefore, insettingtheofficialrate. Unfortunatelyth,eequilibriumrealexchangerateisathwreticalconstructwhichmustbeestimati, not ameasuredquantityforwhichdataexis~ Mor@ver,evenina~ed markegtheempiricalestimationofthe equilibriumrealexchangerateisnoeasytask Estimatingtheequilibriumrateisahighlyinvolvedprocess requiringstrongassumptionsabouttheoperationofthecurrentandcapitalaecounti,aswellastheestimation ofstabletradeandpaymentsrelationshipsovertime15. Estimationofthelong-runequilibriumrd exchangerateforawidesetofcountries,inordertocompare thoseexchangeratestoactualparallelrates,woddgobeyondthelimitedscopeofthispaper. Asaf~ststep towardidentifyingwhereparallelmarketra~s standinrelationtolong-runequilibriumrealexchangerates, lqBmedon~ opti~z~g mo~l oftie paralleml arketexchangerate,MontielandOStrY(1994)~me to muchthesameconclusion.Theyfmdthatinthetransitionbetweensteady-stateequilibriainresponsetoa productivityshock,theparallelmarketpremiummaymovebothaboveandbelowzero,andhenceis“an unreliableindicatorofthesignandmagnitudeofrealexchangeratemisalignment.” lbAnm~r of~proachedhaveb~n triedinthelitera~e. Devarajaran(1995)~es agener~ equilibriummodelthatfactorsinchangesintermsoftradetoestimatetheequilibriumrealexchangerateand theextentofovervaluationintheCFACountrieSB;affesetal (1995)constructaneconometricmodelusinga varietyofmacroeconomictidament,alsforseveralofthesecountries;AhlersandHi.nkle(1995)followa tradeelasticitiesapproachintheirestimation.

23 however,itmakessensetocomparelevelsoftheparallelratetolevelsoftheofficialrate,averagedoverlong periodsof time.Thiscomparisonmaybeinformative,sinceoverstilciently longperiods,thebalanceof paymentsmustonaveragebeatasustainablelevel.Additionallyi,tmaybeusefultocomparethelevelofthe rd officialexchangerateduringperiodswhenexchangeConwolasreh effec4andhenceparallelmarketsexist, toperiodswhenexchangemarketsareunifiedS.uchcomparisonsmayshedlightonthefactorsthatmotivated theimpositionofexchangecontrolsw, hichintummayhaveimplicationsfortherelationshipbetweentheparallel andequilibriumrealexchangeratm. Acomplicatingfactorinusingaveragesofactualexchangeratesasproxiesforequilibriumexchange rates is thatformostcountrim,theprocessof developmentandstructuralchangewillcausethelong-run equilibriumexchangeratetochangeovertime.Inthatsense,alongperiodaverageofactualrealexchangerates mayyiel~atbes~anaverageoflong-runrealequilibriumexchangeratesoverthatperiod.Withthiscaveatin mindhoweverw, estillbelievethatempiricaclomparisonosfactualofficialandparallelexchangeratescanyield usefi insights. First, ouranalysiswillfocusonaveragesofexchangeratesacrossalargesetofdifferentcountria. Therefore,eveniflong-runrealequilibriumexchangeratesfollowparticulartrendsineachindividualcountry, theaveragelong-runrealequilibriumexchangerateforthesampleasawholemaybemorestation~. Second, ouranalysisfm~ uponcomparisonof~mnt typesofexchangerates--officiaalndparallel-duringdifferent regimes–withandwithoutexchangecontrols.Therefore,ourresultswillbe~erable tomisinterpretation mainlyifthetimingofexchangecontrolperiodsinthesamplecoincidewithparticdarmovementsinlong-run equilibriumreaelxchangeratm. Whilewebetieve,asd.iscussedabove,thatexchangecontrolperiodsarelikely tocoincidewithsystematicmovementsinshorf-runequilibriumrealexchangerates, asaresultoftempor~ shockstocapitalflowsorthetermsoftrade,wehavemuchlesscausetobelievethatexchangecontrolshavebeen associatedwithparticulartrendsinZong-runequilibriumrealexchangerates. . Y1 DataandMethodology

24 Tomakethesecomparisonsw, egathereddataontheofficialandparallelexchangeratesforasample of24developincgountri=(SeeAppendix1forthecompletelist). Ofthese,twelvearefromLatinAmerica,ten areAfricanandtwoareSouthAsian.ThesampleisbasedononechosenforanearlierWorldBankresearch projectonthemacroeconomicimplicationsofmultipleexchangerates(seefootnote3above).Thesampleis consideredtobefairlyrepresentative-geographicaallnycwl ithrespecttothealternativepathsofevolutionof theparallelrnarke~Thesampledoesincludemostofthemoreprominentdevelopingcountrieswithsignificant parallelmarkets,outsideeasternandcentralEuropeandtheformerSovietUnion.Thebasicdatasetistheone usedinGheiandKiguel(1992).WeaddedthreeAfricancountries-Algeri~MalawiandSudan;andextended thedatasettothelastquarterof 1994. Tomeanin~y comparelevelsofexchangeratesovertime,wefmt correctedthedataforchangesin prim bydculatingrealexchangerates.Thereareanumberofempiricaldeftitionsofthe realexchangerate (seeHinkleandNsengiyumv(a1995)foradetaileddiscussion)W. eusethebilateralrealexchangeratebetween thecountryweareexaminingandtheUnitedStates(localcurrencyperUSdollar-anincreaseintheexchange rateindicatesdepreciation)T.heconsumerprimindexisusedasaproxyfordomesticprices;theUSproducer priceindexforworldprices. . E.Pus e— (12) P whereeistherealexchangera~, E,thenominalexchangerate,isthe IocdcurrencyvalueofU.S.dollar,Pus isthe USproducerpriceindexandPisthedom~ticCPIlb. lbItisconceivablethattheuseofabilateralexchangerate,usingtheUSproducerpriceindexasaproxy forworldpricesmaybiasourresults,inviewofthesi~lcant movementsoftie USdollarvis-a-visother industrialcountriesduringthe1980s.Resultsofteskforsensitivitywithrespecttochoiceofforeignprice indexarepresentedinAppendix3,forasub-sampleofcountriesfortheperiod1979-94.

25 Thedatausedareendofquarter,fortheperiodfrom1970through199417T.herealofficialexchange rateisindexedsothat1985Q1isequalto100.Therealparallelmarti exchangerateisindexedsothatitsvalue in 1985Q1isequtito 100plusthepremium(inpercent)oftheparallelrateovertheofficialrateinthatbase quarter.Foreachcountry,meanandthemedianaveragevaluesarecalculatedforthefollowing:e,theofficial realexchangeratefortheentireperiod;ep,theparallelrealexchangeratefortheentireperiodofi~existence, thatis,whenexchangewntrolswereineff~t; en.,theofficialnon-tiled realexchangerate,fortheperiods whmexchangecontrolswereineffectandexchangemarketswerenotMled; ande’,theofficialrealexchange rateduringperiods,ifanyexisted,whenexchangemarketiwereunified.Aunificationofexchangemarketsis definedtohave* placeiftheabsolutevalueoftheparallelratedeviateslessthan3perwntfromtheofficial rateforatleast4q~. Thenon-zeronumberistotakeintoaccountmeasuremenetrrors,sincetheoficialand parallelratesarehorndifferentsour= (seetablm). A~S-COUntI’Syummaryofourcalculationsofthefourrealexchangeratecategoriwdescribedabove, basedonmeanaveragesoftheexchangeratedataforeachseparatecountry,isprmentedincolumns1through 4ofTable3. ResultsarepresentedinTable3forthewholesample,aswellastwosubsets.Forthecountri~ ineachsubset,wepresentthemean,median,andstandarddeviationoftherealexchage rate. Incolumns5 though9ofTable3,wecalculmvariousratiosofthedatashownincolumns1through4,andperfoxmbinomial signhts to~e whethertheseratiosdiffersignificantlyfromone.ThePr (Hoistrue) rowindicatesthe probabilitythattheob~ed configuratioonfratioswouldbeobsemed,ifthenu~hypothesis-thattheratiowere equalto1-weretrue. Theregionaglroupingswerechoseninordertotestourhypoth~s concerningthedifferentmotivation andfunctionofparallelmarketsindifferentregions. InLatinAmericaandTurkey,exchangemntrolswere imposed,particularlyinthe1980s,insituationsof~nornic andbalance-of-paymentcsrisisleadingto ITTheoffici~exchangera~s andprim arefromII’vIFI,nternationalFinanciaZStatistic. Thep~allel exchangerat= arefromWorldCurrencyYearbookandInternationalCurrencyAnalysis,Inc.

Z6 strongcapitaloutflows.Inthesecountriwp,ressurmtim capitalflightarelikelytohavecausedtheparallelrate todepreciatewellabovethelong-runequilibriumrealrate,eveniftheofficialexchangera~wmnotespecially overvaluedrelativetoitslong-runequilibriumvalue.tican andSouthAsiancountri~,ontheotherhand, experiencemduchlessmacroeconomidcistress.Inthosecountries,exchangecontrolsweremorelikelytohave arisenasameansofrationingforeignexchangereceiptsinthecontextofaprogressiveovervaluationofthe officialexchangerate. AsdescribedinSectionIV.2,theparallelrateislikelytobemoredepreciatedthanthe long-runequilibriumrateinthismntextaswell,unlessexchangecontrolsarepoorlyenforcedand/orwellenforcedimportbarrierseffectivelycurtailthedemandforforeignexchange. K2 ComparisonsofPeriodAverages Wenowcomparetherealparallelrateeptovariousproxiesoftheequilibriumrealexchangerate. The fmtpossibleproxyfortheequilibriumRERwewnsiderise,theaverageRERforaperiodoftwentyfiveyem. Asmaybeseenincolurnn5,&/e,onavaage,isgreaterthan1,withameanof1.64fortheentifesample.Infact, thereisnocountryinoursampleforwhichtheaveragerealparallelmarketexchangeratewasmoreappreciated thantheaveragerealofficialrate(calculatedforperiodswhentheexchangemarketwasunifiedaswellasnonunified).Totheextentthattheaverageofficialrate,whenaveragedoverastilciently longperiod,isagood proxyfortheequilibriumrealrate,thissugg~~thattheparallelrateisabiasedindicatoroftheequilibriumrate. Howeveru,singtheaverageRERfortheentireperiodmaybemisleadings,inceitinclud~periodswith exchangecontrolsaswellaswhentheexchangemarketisunified.Whenexchangemntrolsareinplace,the nominalpriceofforeignexchangeissetbytheauthorities,andaccesstoforeignexchangeisdeterminedby quantitativerationing.Hence,theofficialrealexchangerateduringperiodsofexchangecontrolislikelytobe moreappreciatedthantheequilibriumrealrate, ItmaybemoreappropriatetousetheRERaveragedover periodsofun~ledexchangemarke~-thatis,em-asaproxyforthelong-runequilibriumRER. Forthewholesample,theparallelrateis,onaverage,moredepreciatedthantheunifiedofficialRER; asindicatedincolumn8,themeanoftheratio,&/e”,is1.43forallwuntries.However,thisresultmash strong

27 @erenm Nw= LatinAmericaandTurkey,withameanof1.58,andMca andAsia,withameanvalueof 1.05(whichisnotsi@lcantly differentfrom1).Hence,totie extentthate“isagoodproxyforthelong-run equilibriumrealexchangera~, epappears,onaverage,tohavebeenclosetothelong-runequilibriumexchange rateinAfricaandAsia butmuchmoredepreciatedthanthelong-runequilibriumlevelinLatinAmericaand Turkey.Thismayseemsurprisingg,iventhefarhigheraverageparallelpretia thathavebeenobservedinmuch of Africa;as shownincolumn7, theratiooftheparallelmarketratetotheofficialrateduringperiodsof exchangecontrolaveraged1.85forAfricaandAsiabutonly1.42forLatinAmericaandTurkey. Thestrong~erenc~ in&/e”,theratiooftheparallelratetotheunifiedofficialrate,betwwnLatin Americaandficaappear toberelatedtoequallymarkeddifferencesinthe evolutionoftheofficialexchange ratebetweenthetworegions.Asindicati incolumn9,inLatinAmericaandTurkey,therealofficialexchmge ratetendedtobemoredepreciatedduringperiodsofexchangewntrolsthanduringperiodswhentheexchange marketsweretiled; themeanratioofe“-/e”was1.12,with9ofthe12countrieshavingratiosgreaterthan1. This is consistentwithour viewthat in Latin Americaand Turkey,unsustainablepoliciesresultedin macroeconomicdisequilibriumw, hich,in turn,triggeredcapiti outflowswhichdepreciatedtheshort-run equilibriumrealexchangerate(arelationshipthatwouldholdinaunifiedexchangemarket)relativetoi~1ongrun level.The governmentdiddepreciatetheofficialexchangerate,butnotby as muchas theshort-run equilibriumratedepreciatedT.herefore,anex~s demandforforeignexchangedevelopedc,ausingtheparallel ratetodepreciateaswell.Asmacrop=urm _ capitaloutflowsmoderatedandreversedthemselvesl,eading toanappreciatioonftheshort-runequilibriumexchangerateandfacilitatingtheunificationofexchangemarkets. IncontrasttoLatin~ca andTurkey,intheMean andSouthAsiancountriesinoursample,thereal officiaelxchangeratetendedtobemoreappreciatedwhenexchangecontrolswereinplacethanwhenexchange marketsw=unified. Asshownincolumn9,for~ca andSouthAsia,theratioof eB9/emwasonly.65,with al15countriesinthisgroupingshowingratioslessthan1. This evidence,whilequ~ledby thelownumberof observationsinthesub-sample,supportsourspeculationthattheemergenceofaparallelmarketinMca and

Lo Asia was typicallytheresultof an appreciationof therealofficialexchangeraterelativem thelong-run equilibriumrate.Authoritieschose,foravarietyofreasons,torationforeignexchangewhilemaintainingan overvaluedrealexchangerate.As theextentof overvaluationinaeased,oftenforeignexchangerationing tightened,andtheparallelmarketgrew,asdidthepremium. ConsideringhowovervaluedtherealofficialexchangeratewasinmanyAfricanandtheSouthAsian countriesr,elativetoitslong-runequilibriumvaiue,itissurprisingthatrealparallelratesinthosecountrieswere notmoredeprwiatedcomparedwithaveragerealofficialexchangeratesduringtheperiodswhenexchange marketswereunified.SectionIV.2showedthat,allelseequal,themoreovervaluedwastheofficialexchange rate, themoreundervaluedwouldbetheparallelraterelativeto thelong-runequilibriumrate. Therefore, unificationpresumablywouldhaverequiredtican andAsiangovernmenttodevaluetheirofficialexchange ratestoaleveltiat wasnotasdepreciatedasthepriorleveloftheparallelrate. Inaddressingthisissue,itisimportanttopointoutthatwehaveaverysmallsampletoexamine-only 5~untriminoursampleinthesubse~AfricaandSouthAsia,unifiedtheirexchangera~. Intwoofthesefive cases(EgyptandT~ania), theratioisgreaterthan1. Additionallya,swepointedoutinSectionIV.2,thereare factorsthatmaylowertheparallerlaterelativetothelevelpredictedbythebasicmodel.Firs~considerthecase whenunderinvoicing+,,ishighbecauseexchangecontrolsarenoteffective-whenenforcemenitslackadaisical andevasionwikprd Th~ forallpracticalpurposes,therelevantratefortheemnomybecomestheparallel rate,whichmay,inthiscase,beclosetotheequilibriumrate. Then,whenexchangemarketsareunified,the officiarlatewouldneedtobedepreciatedtotheleveloftheformerparallelrate,and&/eUwouldbeclosetoone. Ghanaisanexullent illustrationforthis (ChibberandShaffii(1991)). As~nd factorthatmightappreciatetheparallelmarketraterelativetothelevelpredictedbythebasic modelmightbethestrongenforcemt ofimportcontrols.Thiswouldcausethepremiumtobelow,evenifthe officiaelxchangerateismaintainedatasubstantiallyovervaluedlevelcomparedwiththelong-runequilibrium . exchangerate,sinceeffectiveimport~ntrolsreducethedemandforforeignexchangeandtherebyappreciatethe

29 short-runequilibriumexchangerate.Then,if tilcation is associatedwithimportliberalization,thereby depreciating short-runrealequilibriumexchangerate,theofficialexchangeratewillhavetodepreciateby alargeamountifexms demandsforforeignexchangearetobeeliminated.Thisscenariowouldleadtoaratio, &/e”,tiatwouldbelowcomparedwiththepredictionsofthebasicmodeloutlinedinSectionIV.2. Indiaisthe ht examplefmthis-tilcation oftheexchangeratetookplacein1993aspm ofalargerliberalizationeffort, requiring greaterdevaluationoftheofficialexchangeratethanwouldhavekn thecaseintheabsenceof importliberalization. Table3:FullPeriodStatistics,RatiosandBinomialSi~ Tests 1 2 3 4 5 6 7 8 9 Offlciai Parallel Official Official Ratio, Ratio, Ratio Ratio, Ratio, RealER RealER ERwith ERwhen #/e e’”le #/e”” @/e” e-lea Country 1970-94 Par.mkt Unifjed 2fl 3/1 2f3 W4 3/4 ~tin AmericaandTurkey Mean 82.52 121.02 85.13 76.91 1.46 1.03 1.42 1.58 1.12 Median 69.10 115.01 69.43 64.66 1.24 1.06 1.36 1.32 1.10 StdDev 26.17 24.29 47.71 27.28 0.36 0.26 0.34 0.39 0.16 Numbers 1 0 3 0 0 3 Number>1 12 9 12 12 9 Pr(HOistrue) 0.0002 0.019 0.0002 0.0002 0.019 A~“ca&Asia Mem 120.04 222.82 117.94 195.24 1.82 0.98 1.85 1.05 0.65 Median 103.56 186.64 101.56 267.03 1.78 1.00 1.81 0.97 0.57 StdDev 45.99 111.76 45.03 85.81 0.58 0.02 0.58 0018 0.12 Numbers 1 0 12 0 3 5 Number>1 12 0 12 2 0 Pr(HOistrue) 0.0002 0.0002 0.0002 0.500 0.031 All countries Mean 101.28 171.92 101.53 111.71 1.64 1.01 1.64 1.43 0.99 Median 91.38 131.33 91.10 86.24 1.52 1.00 1.54 1.36 1.07 StdDev 41.60 99.81 40.40 74.67 0.50 0.04 0.51 0,38 0.25 Numbers 1 15 0 3 8 Number>1 24 9 24 14 9 Pr(~ istrue) O.000 0.153 0.000 0.006 0.50C, SourceE:S,WorldCurrencYyearbooCku,rrencAynalysIinsc,variouisssuesa,uthor~sculations.

30 K3 TheOficialExchangeRateafierExchangeMarht Unification Theissuesrelatedtounificationcanbetier examinedbyconsideringwhathappenstothe official exchangerate atthattime. Wehavetwentyobservationsofunifications.Insomecases,acountryhastwo episti ofdualexchangerat~, withaperiodofunifiedexchangeratesintheinterim;thesearetreatedastwo separateobservationsS.omecountrimhavenevertiled, andthereforearenotrepresented(seeAppendix2for thecompletelist).Thecompositioonfthetilcation datasetisquitedifferentfromthatusedinTable1.Firstly, threequartersoftheobservationsarehorntheexperienceofunificationinLatinAmerica;onlythreeoutofthe twentyobservationsareforAfrica,withthetwoSouthAsiancountriucompletingthecoun~Theverysmall samplemeansthatthesubsetsresultsaretobeinterpretedwithcautionfortheAfricaandSouthAsiasubset. InTable4,welookatthemean,medianandstandarddeviationofthreevariables:theaverageofficial RERfortheyearbeforethetication (10),theaverageofficialRERintheyearfollowingMlcation (11)and the averageparallelratefortheyearprecedingtheunification(12).Wecalculatetworatios,of theposttilcation officiaRl ERtothepre-unificationRER,egt+l/emmant.dl;theratioofthepostunificationofficial(and sole)RERtothepre-unificationparallelrate,eot+l/&t.Al.sbefore,wereportthemean,medianandstandard deviationforallvariablmaswellasther~ults ofbinomialsigntestfortie ratios. Theresultsareconsiskntwiththefindingsreportedintable3.Theratioofthepost-topre-untication officialexchangerate, ew,+l/em9isi.1a,pproximatelyequaltooneforthewholesample.Again,aggregation obscurathedifferenminresul~betweenthetwosubsefi.Theratio, eut+l/e’gits.l(very)slightlylessthan1for LatinAmericaandTurkey,suggwtingasmallappreciationintheRERfollowingunificationT. hisisconsistent withtheideathancapitalWows resumedasthemacrocrisiswasalleviated,appreciatingtheshort-runreal equilibriumexchangerate andallowingtheauthoritiesto appreciatetherealofflcid exchangeratewhile simultaneousluynifyingtheexchangemti. Theratioisslightlygreaterthan1onaverageforAfricaandAsia, astheexchangeratewasdevaluedfollowingunificationA. gain,thisisconsistentwiththeobservationthatin . threecountria,officiaelxchanger- duringtheexchangewntrolperiodwereovervaluedrelativetotheirlong-

n 31 runequilibriumraks, sothatunificationrequiredthedevaluationoftheofficialrate. Therelationshibpetweenthetild officialrateandthepriorparallelratealsodifferssomewhat thesub-samples.Theratio eO,+l/epct.llearlyislessthan1fortheentiresample,indicatingthatingeneral,the offlcirdrateisnotdepreciatedalltheway@theleveloftheformerparallelrate. WiththeexceptionofGhana andBolivia,theaverageratioislesstian 1forallcountrim.~S rault alsoisunambiguousfortheLatin AmericaandTurkeysubs&Ontheotherhand,berat.io egt+l/&,.dloesnotappeartobesignificantlydifferent fromoneforthe~can andAsiansample,whichisconsistentwithresultsd=cribedabove-averagetiled officialexchangeratesappeartobeclosetotheaverageleveloftheparallelrateinAsiaandMea. - Table 4: UnificationEpisodes-Statistics,RatiosandBinomialSignTests 10 11 12 13 14 OfficialRER, OfficialRER, ParalleRl ER, e8~+1/eaut.1 e“*+l/#~.l Country Year,unif-1 Year,unif+l Year,unif-1 11/10 11/12 - America&Turby Mean 85.15 85.62 109.31 1.00 0.83 Median 81.99 75.74 101.98 0.98 0.87 StdDev 29.89 33.19 54.88 0.09 0.15 Numbers 1 9 15 Number>1 7 1 Pr(~ istrue) 0.576 0.001 Aftia &Ash Mean 183.67 201.5~ 203.81 1.08 0.98 Median 172.71 155.21 180.99 1.11 0.94 StciDev 71.92 87.56 85.81 0.12 0.13 Numbers 1 1 4 Number>1 4 1 Pr(HOistrue) 0.188 0.188 All countries Mean 108.61 113.21 131.81 1.02 0.86 Median 99.42 83.77 118.08 1.00 0.90 StdDev 60.60 71.42 75.28 0.11 0.16 Numbers 1 10 19 Number>1 11 2 Pr(~ istrue) 0.500 0.0001 SourceIF:SW, orldCmencyYearbooCku,rrenc~yysis Lncv,ariouisssuesa,uthorcsalcu~ons.

32 VI.Conclusion Wenows~ themostimportantfindingspresentedinourpaper.Tobeginwith,ourtheoretical analysisindicatedthatwhentheemergenceofaparallelexchangemarketismotivatedbytheovervaluationof theofficiaelxchangeraterelativetothelong-runequilibriumred exchangerate,theparallelmarketrateislikely, onaveraget,obemore@reciatedthanthelong-runequilibriumrealextiangerate. Thegapbetweentheparallel rate andthelong-runequilibriumrate is likelyto be smaller,to theextentthatexportreceiptsurrender requiremen~arenutwelldor~ andtotheextentthatbarrierstoimportsandothercommercialpoliciesthat tendtoappreciatetheshort-runequilibriumrealexchangerate arewell-enforced. Moreovero,urtheoreticalanalysisindicatedthateveniftie officialexchangerateissetati~long-run equilibriumlevel,aparallelmar~ mayariseinordertomeetthedemandsofresidentsseekingtoaugmenttheir holdingsofforeignassets.Duringtheperiodwhenforeignassetsarebeingaccumulated--thaits,whencapital mghtis~g-the parallelexchangeratewillbemoredepreciatedthanitsownequilibriumvalue,andhence probablymoredepreciatetdhanthelong-runequilibriumextige~ fmtheeconomyasawhole.Additionally, becausetheparallelmarketrateisanasw price,andexhibitsthevolatilitythatischaracteristicofallassetprices, thevalueoftheparallelrateatanygivenmomentislikelytobeaparticularlypoorindicatorofthelong-run equilibriumexchangerate. Th~econsiderationssuggestthatonbalance,theparallelrateislikelytobemoredepreciatedthanthe long-runequilibriumrealexchangerate,andhencetheofficialexchangerateinaunifiedexchangemarketwill ingenerahlt besetatalevelthatismoreappreciatedthanthepriorparallelrate(averagedoverasuitablylong period).Inthisp~, wedidnotcomparea parallelexchangeratestoestimatesofthelong-runequilibrium rateindifferentcountries,owingtothed~lcultyofestimatingequilibriumratesforalargesample.However, wecomparedmulti-yeaarveragaofrealparalle~l torealofficiarlatesinasampleof24developingcountries, andmadeanumberofuseti findings. First, we foundthatforthesampleas a whole,therealpartiel marketratewasuniformlymore

33 depreciatedthantheofficialexchangerate,evenwhentheofficialratewasmeasuredonlyduringperiodswhen theexchangemarketswereunified.Duringperiodswhentheexchangemarketisunified,andthereareno exchangecontrolstobridgethegapbetweensuppliesanddemandsforforeignexchange,theofficialexchange rateismorelikelytobeclosetotielong-runequilibriumrateonaverage.Hence,thisrepresen~partialevidence thattheparallelrateten~ tobeundervaluedrelativetothelong-runequilibriumexchangerate. Secondwefoundimportandtifferencesintherelationshipbetweenparallelandofficialexchangerates amongdifferentsubsefiofourcountrysample. InLatinAmericaandinTurkey,theemergenceofparallel exchangemarketsappears~ havereflectedasharpdepreciationoftheshort-runequilibriumexchangerate relativetoitslong-runvalue,nottheappreciationoftheofflciaiexchangeratehornitslong-runequilibrium value.Inthose~untri~, theparallerlatewasclmlydepreciatedcomparedwiththeofficialexchangerateduring periodswhentheexchangemarketswereunified.However,theundervaluednatureoftheparallelratedoesnot ap~ tOhavereflect~ fie overvaluat.ioonftheofficialexchangeraterelativetothelong-runequilibriumreal exchangerateduringperiodsofexchangecontrols,sincetheofficialexchangerateinthissubsetwasonaverage moredepreciatedduringtheperiodswhenexchangecontrolsw=eineffectthanintheperiodswhenmarketswere unified. Wesurmisethatacombinationofinternalandexternalshocksledtomacrmnomic turbulenceand capitalflightinLatinAmericaandTurkey,mainlyinthe1980s.Thesedevelopmentsi,nturn,depreciatedthe short-runequilibriumrealexchangeraterelativetoitslong-runlevel.Whiletheauthoritiesdepreciatetheir officialexchange~-possibly eventolevelsmoredepreciatedthanthelong-runequilibriumrate--theydidnot dosobyenought.or~olveex~s demandsforforeignexchange.Thatis,evenifofflcialexchangeratesduring theexchangewntrolperiodwereundervaluedrelativetothelong-runequilibriumrate,theywereovemalued relativeto theshort-runequilibriumrate,therebygivingrisetoparallelexchangemarkets. Asaresulgthe parallerla~ probablywereeva moreundervaluerdelativetothelong-runequilibriumratethanweretheofficial rates.

34 Third,wefoundthattheAfricanandSouthAsia countriesinoursamplebetterfitourpreconception thattheemergenceofparallelexchangemarketsreflectstheovervaluationoftheofficialexchangeraterelative toitslong-runequilibriumvalue.Amongthefewcou.ntriminthissubsetthatexperiencedperiodsoftiled exchangerates,therealofficialexchangerateclearlywasmoreappreciatedduringperiodswhenexchange controlswereineffwtthaninperiodswhenexchangemarketswereunified.Thissuggeststhatincontrasttothe LatinAmericaandTurkeycase,therealappreciatioonfofficialexchangeratesinAfticaandSouthAsia,relative tolong-runequilibriumvalues,wasthemainfactorunderlyingtheemergenceofparallelmarkets. Ontheotherhand,inAfricaandSouthAsia theparallelexchangeratewasnotsi~lcantly more depreciatedthantheofficialexchangerateduringperiodswhenmarketswereunified--putanotherway,when exchangemarketswereunified,theauthoritieshadtodepreciatetheofficialrateallthewaytothelevelofthe formerparallelrate. Therelativesimilarityofparallelandunifiedofficialexchangeratesunderscor~thefact thatinthecaseofsomeMean countries,exchangecontrolsmayhavebeensopoorlyenforcedthattheparallel rateeff~vely mimickedtheroleoftheofficialrateinaunifiedexchangemarke~Additionallyi,nsomeAsian countries,well-enforcedimportbarriersconstrainedthedemandforforeignexchangewhenexchangecontrols wereineffec~therebyappreciatingboththeshort-runequilibriumexchangerateandtheparallelrate;exchange marketsweretiled ataboutthesametimeasimportbarrierswerelowered,makingitnecessarytodepreciate theofficialexchangeratesubstantiallyinordertomaintainthebalan~ ofpaymentsinaunifiedmarke~ OurfindingsforsomeWean andAsiancountriessuggestthatinsomecases,theparallelratemight notbeaninappropritiproxyforthelong-run~uilibriumexchangerate,andhencecouldbeaguidetosetting theofficialrate. Somewhatspecialfactorswereoperativeinthesecountrieshowever--verypoorlyenforced exchangecontrolsinsomeAfricancountri~,verywell-enforcedimportcontrolsinsomeAsiancountries.More gen~y, theparallelmarketratewouldseemtorepresentanupperbound,intermsoflocalcurrencyperdollar, ontheappropriateleveloftheofficialexchangerate.

35 REFERENCES: Agenor,Pierre-Richard(1992),parallelCurrencyMarketsinDevelopingCountries:~eory, Evi&nce,and PolicyImplications,PrincetonEssaysinIn@mationaFlinanceNo.188,Princeton,N.J. AhlersT.O.,andL.E.He (1995),“Estimatingtheequilibriumrealexchangerate:anupdatedresourcebalance target/tradeelasticitiesapproach”,WorldBankWashingtonDC,fotihcoming. Bties, J.,I.A.ElbadawiandS.A.O’Connell(1995),“Econometric~timatesoftheequilibriumRER”,World Bank,WashingtonDC,fo~hcoming. ChibberA, , andN.S- (1991),“Extige ref~ parallelmarti andinflationinWca: thecaseofGhana”, inA.CribberandS.Fischer,eds,EconomicRefom inSub-Sabran Afi”ca,WorldB& WashingtonDC. Devaraju S.(1995),“EstimationofRERmisalignmenutsingthe1-2-3model”,WorldB* WashingtonDC, forthcoming. Dombusch,R.,D.V.Dantos,C.Pechman,R.R RochaandD.Simoes(1983),“Theblackmarketfordollarsin Brazil”,QuarterlyJournalofEconomics,Vol98,pp.25-40. Ghei,N., andM.A.Kiguel,(1992),“Dualandmultipleexchangeratesystemsindevelopingcountrim:some empiricalevidence”,PolicyResearchWortingPaperNumber881,TheWorldB* WashingtonDC. and S.A.O’Connell(1996),“ParWelexchangeratesindevelopingcountrim:lessonstim eightw’estudies”inA4.AKiguel,J.S.h“zondoandS.A O’Connell,e&,ParallelExchangeRatesinDeveloping Countries,MacMillanb,ndon andSt.Martin’sPress,NewYork. Hinkle,L.E., andF. Nsengiyumva(1995),“Realexchangerateindexes:a user’sguideto theirtheoretical definition,empiricalmeasuremenatndinterpretadon”W, orldB*, WashingtonDC,processed,May8,1995. Kamin,S.B.(1995),“Contractionarydevaluationswithblackmarketsforforeignexchange”,JournalofPolicy Modeling,Vol.17,number1,pp.39-57 Karnin,S.B.(1993),“Devaluation,exchangecontrolsandblackmarketsforforeignexchangeindeveloping countri~,JournalofDevelopmentkonomics, Vol40$pp.151-169. Kiguel, M.A.,andS.A.O’Comell(1995),“Parallelexchangeratesindevelopingmuntries”, WorZdBank ResearchObserver,Vol10,numbr 1,pp21-52. Montiel,PeterJ.andJonathanD.Ostry(1994), “Theparallelmarketpremium:isitareliableindicawrofreal exchangeratemisalignmenitndevelopingcountries?”IMFStaffPapers41,pp.55-75. MontielP,eterJ.(1996),“Thetheoryofthelong-runequilibriumrealexchangerate,”WorldB* Washington, DC,forthcoming. NOW*M.(1984),“Quantitativmentrolsandunofficiaml arketsforforeignexchange”I,MFStaffPapers31,pp. 404-31 She~ M.A(1976),“Blackmarketforforeignexchange,capitalflowsandsmuggling”J,ournalofDevelopment Economics,Vol3,pp.9-26.

i * m .

I I 37 Appendix2 Table2.A:UnificationEpisodes country 10 11 12 13 14 e(nu,t-l)OfficialRERe(qt+l)O~ial RER e(p,t-l)ParallelRER e(u,t+l)/e(nu,t-1) e(u,t+l)/e(p,t-1) Year,Unification-1Year,Unifi-tion+l Year,Unification-1 11/10 11/12 Mean IMedian Mean Median Mean Median Mean Median Mean Median LQtinAmerica and Turkey Argentina 76.45 69.89 74.17 75.55 147.95 141.42 0.97 1.08 0.50 0.53 Argentina 26.52 26.41 26.03 25.93 29.40 29.04 0.98 0.98 0.89 0.89 Bolivia 126.58 126.10 136.51 136.50 135.79 134.01 1.08 1.08 1.01 1.02 Bolivia 125.96 126.17 128.00 127.70 134.21 134.36 1.02 1.01 0.95 0.95 B- 60.01 59.92 59.30 59.08 64.74 65.17 0.99 0.99 0.92 0.91 Chile 58.23 57.82 64.18 64.31 65.00 65.32 1.10 1.11 0.99 0.98 Colombia 99.42 99.68 83.77 84.26 104.46 104.28 0.84 0.85 0.80 0.81 Dom Rep 57.21 56.98 56.12 56.36 61.18 60.92 0.98 0.99 0.92 0.93 Ecuador 126.11 126.18 126.39 127.84 141.07 140.06 1.00 1.01 0.90 0.91 Mexico 103.69 110.30 106.98 106.10 114.40 115.15 1.03 0.96 0.94 0.92 Mexico 87.53 88.01 76.20 75.50 92.51 93.29 0.87 0.86 0.82 0.81 Pem 64.50 63.99 75.28 75.01 132.37 146.55 1.17 1.17 0.57 0.51 Urumay 68.78 68.40 67.09 67.18 80.93 81.16 0.98 0.98 0.83 0.83 Venezuela 128.77 108.14 149.58 149.64 277.38 272.90 1.16 1.38 0.54 0.55 Turkey 64.49 65.59 57.84 58.19 68.11 68.22 0.90 0.89 0.85 0.85 Turkey 88.19 87.38 82.40 83.32 99.50 97.95 0.93 0.95 0.83 0.85 Afi”caandAsia Egypt 172.71 171.70 155.21 156.55 180.99 176.55 0.90 0.91 0.86 0.89 Ghana 212.66 212.93 266.21 271.59 217.60 223.52 1.25 1.28 1.22 1.22 Tanzania 305.99 306.85 340.09 334.37 361.34 355.90 1.11 1.09 0.94 0.94 India 120.01 119.88 138.24 138.99 141.04 140.95 1.15 1.16 0.98 0.99 Pakistan 106.97 106.40 107.73 108.31 118.08 116.50 1.01 1.02 0.91 0.93 LatinAmerica&Turkey Mean 85.15 83.81 85.62 85.78 109.31 109.36 1.00 1.02 0.83 0.83 Median 81.99 78.63 75.74 75.52 101.98 101.12 0.98 0.99 0.87 0.87 StdDev 29.89 28.91 33.19 33.22 54.88 54.17 0.09 0.13 0.15 0.15 AMca & Asia Mean 183.67 183.55 201.50 201.96 203.81 202.68 1.08 1.09 0.98 0.99 Median 172.71 171.70 155.21 156.55 180.99 176.55 1.11 1.09 0.94 0.94 ,StdDev 71.92 72.41 87.56 86.23 85.81 84.65 0.12 0.12 0.13 0.12 All countries Mean 108.61 107.56 113.21 113.44 131.81 131.58 1.02 1.04 0.86 0.87 Median 99.42 99.68 83.77 84.26 118.08 116.50 1.00 1.01 0.90 0.91 StdDev 60.60 60.74 71.42 71.13 75.28 74.31 (-)-11 0.13 0.16 0.16

38 Appendix3:SensitivityofResultstochoiceofRealExchangeRate ThetrendsinsectionVareestimatedusingtheUSproducerpriceindexasaproxyforworldprices. Thismaybiastheresul~,particularlyfortimeperiodswhenmovementsintheUSrealexchangeratediverge significantlyfromthoseofotherindustrialcountries,astheydidovermuchofthe1980s.Weestimatethe sameratiosasinTable3,usingtrade-weightedmultilaterarlealexchangerat= (seeTables3.Aand3.B below). ThedatawereavailableintheIn&mationaFlinancialStatisticsdatabase,oftheIMF, foron.lytenof thetwenty-fourmuntriesinoursample,forthetimeperiod1979to1994.Whileitwouldbepossibleto calculatemuldlateralrealexchangeratesforallthecountriesinthesample,fortheperiodunder considerationt,hiswoddbealaboriouspr~s. Further,thisdatasetcoversashortertimeperiodthanthe oneusedinthestudy(1970to1994).However,therealUS$bilateralratewouldhavedivergedsi~lcantly hornthetradeweightedmultilaterarlateprimarilyinthe1980s;thisdatasetincludestheperiodofinterest. Therefore,weusethesmallerdatasetreadilyavailabletocomparethevaluesobtainedfortie ratioswe calculateusingthebilateralrealexchangeratewiththoseforavailabletradeweighted,multilaterarleal exchangeratm.. Theuseofmultilateraelxchangeratesyieldestimatesoftheratiosunderconsiderationthatare remarkablysimilartothoseobtainedusingbilateralresults.Theoneresdt thatisdifferentisthebinomial signt~t fortheratio,eDa/e’T.heprobabilitythatHoistrueismuchhigherwhenthemultilaterarlateisused. Thisimprobablyduetotheverysmallsample:just fiveobsemationsT. heordersofmagnitudedonotdiffer markedlyforthemeasuresofcenti tendency. Thechoiceofabilateralrealexchangeratewouldnotseemtoimpactontheresultsinanysignificant way.Atthesametime,usingthebilateralrealexchangeratehasotheradvantagesI.nparticular-,usingthe bila~ralratepermitsanalysisforalargersampleofcountries,foralongertimeperiod.However,thereare

39 someinstanceswherethedifferencesinthevaluesusingthedifferentrealexchangeraw aremoretian triviaI (seeTable3.B,particularlyUruguay(columns8and9,andNigeria).Iftheanalysisistobedoneforasingle country,theuseofthetrade-weightedmultilaterarlealexchangeratewouldbepreferable.Forasampleas largeastheoneinthisstudy,theadvantagesofusingthebilateralrealexchangera~ aregreater. A uuab 4u=* uvaanpua ann~ -mmutwa us Uamu ATAUS*SSUC - - -- —.--- .--.= - .—.— 1 2 3 4 5 6 7 8 9 mlciai Parallel mlcial omtid Ratio Ratio, Ratio, Ratio, Ratio, RealER RealER ER ERwhen #le e“’le &/e”” eple” e“’/e- W/ 1979-94 Parmkt -led 08 Variable e 4 e e“ 2/1 3/1 2/3 2/4 3/4 SummaryStatistim Wean Bilateral 118.09 191.01 118.69 8’7.68 1.55 1.01 1.55 1.62 1.19 Multilateral 175.73 279.88 172.69 149.36 1.53 1.00 1.56 1.50 1.lC Median Bila~ral 106.28 168.57 104.21 72.18 1.36 1.00 1.36 1.60 1.19 Multilawral 132.80 186.09 124.66 103.88 1.38 1.00 1.40 1.54 1.12 StandardDeviation Bilateral 54.66 116.71 54.10 35.74 0.37 0.06 0.42 0.25 0.31 Multilateral 120.95 230.58 117.96 104.42 0.37 0.07 0.43 0.30 0.33 BinomialSignTests Numbe<r=1 Bilateral o 6 0 0 2 Multilateral o 8 0 1 3 Number>1 Bilateral 10 4 10 6 4 Multilateral 10 2 10 5 3 Pr(HOistrue) Bilateral 0.0001 0.377 0.0001 0.0156 0.3438 Multilateral 0.0001 0.055 0.0001 0.1094 0.6563 ——.- — — . . Source:M-SW, orldCurrencyY*Q CurrencyAnalysis,Inc,variousissues,authorscalcula&ions.

c I —

i 1 A [0ml lb C661 Ii)ZM[ lb 1661 . * Ii)0661 8 ! lb686[ 8 .“ lb6861 lb8S6[ t * lb8861 1 [bL861 1 0 lbL861 . lo986[ .,’ .- . l . - . - . - . - ... . -8. . . - . -o - . ... - - --- .2 -- ..~- Z lb S861 * * * * . l 1 b 09 S 8 8 6 6 1 1 4 - “ o - -- - - -- - . - - : .- . - * - - a ---- — l5 l l b b ~ C 6 8 1 61 - . ..“s # =---—- - 4, lbZ861 l * :l Ib 1861 . . i [b086[ : l * i lb9L61 , * 10SL61 * . . -., ..-..--- .----------” .l lbVL61 “. , * + - - - ==-. . . - --: 2l [0SL61 ., “*- . . - - ‘ 8 . . * I1 8 l* ‘. # l l b b Z IM L6 1 1 ‘. I lboL61 Ibwl 10*[ ( IGC661 l lbf661 m [bz661 10Z66I [b1661 $ / 101661 s ..........---.-----. - . . - - - ,.-- lbwI .e....-.Q ----- . . . . . - -- - - 10066[ -- .9------------ 1 lb6861 “ .. .-. [06861 l. l. “. “. .lb8861 “. * lb8861 s .“ l. lbL86[ *. l . . . - . .“ , * . * --” IbL861 , . 4 Ib9861 . .l [2)9861 -- . - . . . # l l b bW S8 6 6 1 [ ..* l l . # .* -* [0s861 -. . . . lbH6i -.. *-C l . . i“1-. . . 1 [0 2) Z C 8 8 6 6 1 1 “.. ..-.. l .* . 1 l l b b z ~ 8 6 6 1 1 =“. . lb 1861 #. 8“ lblS61 8 lb086[ lb0861 lb6L61 l: lb6L61 lb8L61 .l* * I08L61 . ....-. .-“” [0 LL61 a’ lbLL6[ -. - _ - - -9-.. . 7lb9L61 c:a Ib9L61 . “. . Ilb sL61 “ . . “ Ibsul ,-- , l , * ’ 1 lb 0t C L6 L6 1 1 : * 1 [ t lotL61 * -l.’ lbZul .“ . ‘., * i t [0 IL61 8 . i IOIL61 1 I I %, [2)OL61 I I 1 4 10OL61 o 8Z 8:” mm N

IOMI -o--- IZ)M61 -...-- . l lbf661 ‘----------- [05661 10266[ l,... lbZ661 .’ 101661 t*. la I&s[ , 100661 lbWI l s: ,.. .’ lb6861 l .* lb6nl 108861 ‘. -$ lbm61 . “. , * 9 lbL861 l* lbL861 : Ib9861 lb986[ “. . ..’ ‘* lbs861 c:’ [i)S861 .,“ > lbW61 . l “. “ . lb *I --”. .. . --””” lbc861 “==,.. * lbU61 .-. lbZ86[ l* Q o 12)Z861 [01861 !0 1861 . . 100861 lb086[ 8’ Ioul 8 l lb 6L61 .* lb8L61 . l lbKbl lbu.bl l * lbml .0 1b9L61 8 ,’ lb9UI lbsL61 :’ 1?)Su; 9* lb*L6[ ---l--* lbVu[ lofL61 #- . .- [0 ml .1 IbzL61 8 .** lbml l I b b l o f l 6 . 1 61 1 < 8 l. . - . . . . . . . . . . . . [b IL61 I 1 I . [0?661 , [z)ml . lbf661 e . l.0 lb f661 - . ~ .* l . . l [ b 02 1 6 6 6 6 1 1 “ 9 . - : ” * -* 1 [0 01 Z 66 6 1 61 %= . ., lbo661 =., lb ,. ----- . .. l. .: [0686[ %8O .’> . lb 0 68 6 6 6 1 [ “-. . . . --- -.. -. l. l. l [ b b L 8 8 8 6 6 1 [ l. l- . l . . . . , - . . l 1 b 08 L 8 8 6 6 1 1 *. ‘. l lb9861 ,.* 109861 .“ .- -’. c: lbS861 # 10ml 9* *’ ,. - .0 * lbW61 l l . lbml *., “- . ---- 10t861 . . l ‘ . . 0 * lb5861 -. l . . s l [ a 0 Z [8 8 6 6 1 1 ... -l l. :. 1 lb 0Z 1 n 86 6 [ 1 . .* 9 * lb0861 --- .. 100861 . . a [0 6161 ** :. 10al 9 lb8L61 l. 8 lb8L61 9 ,* “ .’ lbU61 ,.* lb&l l # lb9UI lb9UI lbs161 .- l [bSUI l lb bc *L L 61 61 %. .’ 10tf61 . : -.’ lbEf61 l * lb Zf61 e:.: - . “ * a 10Z&sl .“ , 10IL61 ,.- l * [bIL61 .: [0of61 I h —

l Peru Turkey !50 120 * ,88 # !00 100 O ,. . ,- # , * 9 l-l .- L.L , , , s 9 80 50 * 8 I 9 60 00 l = * @ l u ,“ -..** .*S l . 40 D ,.. “.c ,, .,- . , , 50 l 20 0 I—Official -----0ParalleIl I—Official ------ ParalleIl I I Uruguay Venezuela 140I 350 , l I ,* I .* 100 250 80 200 60 150 40 100 20 50 t I—Official ------ ParalleIl I—Official ------ ParalleIl

,-- 2/ , 0--.., ,l :s ‘.- l :’ - [ 1 0 0M C6 l 61 . :* .s . . . .....---, 10266[ “.* l. * . - 101661 . ,. l -.“ lbwl s. l l l. l .* l9 l , l. ,.-. ) 1 l l b b 08 L 6 8 8 8 6 6 6 1 1 [ ,-@ ll * l . l : * , 9 * ,---- ... l .l 1 lb9861 ., * ‘* ‘a .> [0 S861 ‘. b S* ll. . IbT861 .* e -., - .l.-, lb t861 -*, #*. “ . &#45;&#45;&#45; . &#45;&#45;&#45; . -- . . . . . . . . - -- =“. . . *. ... lbZ861 *.’-. . .. . ‘ .l . . * . “ * * 10[861 . . -“ 9= * . - - - .* -’t: lb0861 . .8 . lbUI ,- .. . . - - - .* lb8MI . .9* “. l. lbUI 0% - . . . . . - . - . . . . . - . - . . . . . - . - ., l l. . .. . . 9. lb9L61 ,-. -. l. 1.* lbSL61 -- =,--D 8 9’ lb*L61 ., l 10EL6[ .: * .’ lbZL61 -.= lb IL61 # l I Ibou[ 8888888” *Qln*m~. — [bWI 12)f661 lbZ6I6 . c“ 101661 ,.-.““ “=-. . - - . . . . .. l. lb0661 9‘8 - . .--, lb 6861 ...** l . lb886[ l lbL861 109861 [0 S861 ‘.l -. [bH61 l. . l , : Ii)f861 : .’ / * l.l.* lb2861 - =,. # “.O lb 1861 l . . . ,. .8 .“ 9 8 * l ll l . * 1 1 l lb b 0 06 U 0 8 L 8 L 6 b 6 6 1 i 1 [ ._-. -. -. . . . -.- - .- . - * * - . ‘ - - - . - * ..’ .. * : . l , - ., ..* lb9L61 ,----- -- - . . - - - ., “=. >. l # =. -. .0-.l 10SL61 et. . . .. *-.-* Q.l.* [i)tL61 ..* : * ‘. * .* l I b ZI f m L6 l 1 l. - = ,. . , . *. -. . , . , = lb IL61 .’ .. ... .. a--- IboL61

. . --= 4 l ---- .“’ l -- --~ = l-’- :. ,... .. .?- ..9 .:-D -. .0l .a. .- ..- . .-Q. - * .: = l ,- . -- . . . . . . - ,, “ - - . - - - . .. . . . : 8- ‘. c . .= -.. o-- :. --- -- . -..-.-- . ‘. < 1 l. “. *l ‘ =--- . . --- - ------ . . . -*- ..- ------ , l1’ , *-.=.-: --- -. ..--”.. 1. . .#. .. . . . . . “. . 8- . . ~ . : - =--“-=- :. ,. -“ .’ l ..* l.., . 8“-:.- -’ .- - - . - . . . ,.” B # l.. “ . ., .“ l...-::% “ =., .:* .-* -- - - - . . . ‘. ,.. # .“ “e :. m l . -’ l8 .8 .“ :. #.“ ‘. I I 1 I { I 8 — , IZ)WI ,. l ‘. lbc661 12)266[ .- . c ‘-. [0[661 .’ .- .“ 9 .. 100661 *“ . l* .- lb6u61 .+. l~ . .. . . . . - - . - - - -“ . . --- ”--9 “.- IbS861 l- - =. . . . -. “* It)&l .“ l. --.” ...=.- . . lb9n61 l * Ibm[ 9’ , Ibml ~.- : : lbE861 l ‘> lbzs61 : “. 8 . 101861 .. l# , . 100S61 -. . . .” lo61bl #.’ 1 .’ . -“. -. . . 1 - - . - # 1 lo 0 8f L 6 L 1 6[ l , : , ’ * ‘., 8 # lb9L61 8 8 , .*- :* 10SL61 ,“ # 10*UI ‘,* ------ a .l t : - : ...- : l.- . a - lb C1.lsl 8 l 8 8 , “-=. =. , .“ o

47 International Finance Discussion Papers IFDP Number Titles Author(s) 1996 564 TheUseof theParallelMarketRateasa Guide NitaGhei to SettingtheOfficialExchangeRate StevenB. Kamin 563 CountryFundDiscountsandtheMexicanCrisisof Jefiey A. Frankel December1994: DidLocalResidentsTurn SergioL. Schmukler PessimisticBeforeInternationalInvestors? 562 EasternEuropeanExportPerformanceduring NathanSheets theTransition SimonaBoata 561 Inflation-AdjustedPotentialOutput JaneT. Haltmaier 560 TheManagementof FinancialRisksatGerman AllenB. Frankel NonfinancialFirms: TheCaseof Metallgesellschafi DavidE. Palmer 559 BroadMoneyDemandandFinancialLiberalization NeilR. Ericsson inGreece SunilSharma 558 StockholdingBehaviorof U.S. Households:Evidence CarolC.Bertaut fromthe 1983-89Surveyof ConsumerFinances 557 Firm SizeandtheImpactof Profit-MarginUncertainty VivekGhosal onInvestment: Do FinancingConstraintsPlaya Role? PrakashLoungani 556 RegulationandtheCostof CapitalinJapan: A Case JohnAmmer Study MichaelS. Gibson 555 TheSovereigntyOption: TheQuebecReferendumand MichaelP. Leahy MarketViewsontheCanadianDollar CharlesP. Thomas 554 RealExchangeRatesandInflationinExchange-Rate StevenB. Kamin BasedStabilizations:AnEmpiricalExamination 553 MacroeconomicStateVariablesasDeterminants JohnAmmer ofAssetPriceCovariances 552 TheTequilaEffect: TheoryandEvidencefrom MartinUribe Argentina 551 TheAccumulationof HumanCat)ital: Alternative MuratF. Iyigun MethodsandWhyTheyMatter ‘ AnnL. Owen Pleaseaddressrequestsfor copiesto InternationalFinanceDiscussionPapers,Divisionof InternationalFinance,Stop24,Boardof GovernorsoftheFederalReserveSystem, Washington,D.C. 20551. \

48 International Finance Discussion Papers IFDP Number Titles Author(s) 1996 550 AlternativesinHumanCapitalAccumulation: MuratF. Iyigun Implicationsfor EconomicGrowth AnnL. Owen 549 MoreEvidenceontheLinkbetweenBank MichaelS.Gibson HealthandInvestmentinJapan 548 TheSyndromeof Exchange-Rate-Based EnriqueG. Mendoza StabilizationandtheUncertainDurationof MartinUribe CurrencyPegs 547 GermanUnification: WhatHaveWeLearned JosephE. Gagnon fromMulti-CountryModels? PaulR.Masson WarwickJ. McKibbin 546 Returnsto ScaleinU.S.Production: Estimates SusantoBasu andImplications JohnG. Femald 545 Mexico’sBalance-of-PaymentsCrisis: A Chronicle GuillermoA. Calvo ofDeathForetold 544 TheTwinCrises: TheCausesofBankingand GracielaL.Kaminsky Balance-of-PaymentsProblems CarmenM.Reinhart 543 HighRealInterestRatesintheAfierrnathof GracielaL. Kaminsky Disinflation: Is itaLackof Credibility? LeonardoLeiderrnan 542 PrecautionaryPortfolioBehaviorfroma Life-Cycle CarolC.Bertaut Perspective MichaelHaliassos 541 UsingOptionsPricesto InferPDF’sforAssetPrices: WilliamR.Melick AnApplicationto OilPricesDuringtheGulfCrisis CharlesP. Thomas 540 MonetaryPolicyintheEnd-Gameto Exchange-Rate StevenB.Kamin BasedStabilizations:TheCaseof Mexico JohnH. Rogers 539 ComparingtheWelfareCostsandtheInitialDynamics MartinUribe ofAlternativeTemporaryStabilizationPolicies 538 LongMemoryinInflationExpectations:Evidence JosephE. Gagnon fromInternationalFinancialMarkets

49 International Finance Discussion Papers IFDP Number Titles Author(s) 1996 537 UsingMeasuresof Expectationsto Identi@the AllanD. Brunner Effectsof a MonetaryPolicyShock 536 RegimeSwitchingintheDynamicRelationship ChanHuh betweentheFederalFundsRateandInnovationsin NonborrowedReserves 535 TheRisksandImplicationsof ExternalFinancial EdwinM.Truman Shocks: LessonsfromMexico 534 CurrencyCrashesinEmergingMarkets: An JeffreyA. Frankel EmpiricalTreatment AndrewK. Rose 533 RegionalPatternsintheLawof OnePrice: The CharlesEngel RolesofGeographyVs.Currencies JohnH. Rogers