German Unification: What Have We Learned from Multi-Country Models?

BoardofGovernors oftheFederal ReserveSystem International FinanceDiscussionPapers Number547 April 1996 GERMANUNIFICATION: WHATHAVEWELEARNEDFROM MULTI-COUNTRYMODELS? JosephE.Gagnon,Paul R.Masson,andWarwickJ.McKibbin NOTE: International FinanceDiscussionPapers arepreliminarymaterials circulatedtostimulate discussionandcritical comment. References inpublications toInternational FinanceDiscussion Papers (otherthananacknowledgment that thewriterhas hadaccess tounpublishedmaterial) shouldbeclearedwiththeauthororauthors. 03/25/2 - R:\DOC\AI\PRM\DOC\GERMANY\GERMANY.DOC

DATA ENTRY FORM: AWPE PD April 1996 DATE (MONTH) (DAY) (YEAR) TI German Unification: What Have We Learned from Multi-Country Models? SUBJECT/TITLE AU Gagnon Joseph E. AUTHOR(S) (LAST NAME) (FIRST NAME) (MIDDLE INITIAL OR NAME) Masson Paul R. McKibbin Warwick J. AA Gagnon: Division of International Finance, Board of Governors of the Federal AUTHOR'S AFFILIATION Reserve System; Masson: International Monetary Fund; McKibbin: Australian National University and The Brookings Institution Board of Governors of the Federal Reserve System SR International Finance Discussion Papers : 547 SERIES NAME REPORT NUMBER Division of International Finance SERIES ADDRESS Board of Governors of the Federal Reserve System, Washington, DC 20551 PG 47 PAGES PR no charge PRICE JE E63, E65, F36, F47 JOURNAL OF ECONOMIC LITERATURE CLASSIFICATION NUMBER KW Exchange Rate Mechanism, Germany, rational expectations, transformation KEY WORDS AB ABSTRACT (Abstract follows.)

ABSTRACT This studyreports onearlysimulations oftheeffects ofGermanunificationusingthree different rational-expectations multi-countrymodels. Despitesignificant differences intheir structures andintheimplementations oftheunificationshock,themodels deliveredanumberof commonresults that provedtobeareasonablyaccurateguidetothedirectionandmagnitudeof theeffects ofunificationonmost keymacroeconomicvariables. Inparticular,unificationwas expectedtogiverisetoanincreaseinGermanaggregatedemandthat wouldput upwardpressure onoutput,inflation,andtheexchangerate,anddownwardpressureonthecurrent account balanceinGermany. Themodel simulations alsohighlightedthecontractionaryeffects ofhigh Germaninterest rates onothermembercountries oftheExchangeRateMechanism ofthe EuropeanMonetarySystem.

German Unification: What Have We Learned from Multi-Country Models? JosephE.Gagnon,Paul R.Masson,andWarwickJ.McKibbin1 I.Introduction and Goal of thePaper TheunificationofEast andWest Germanygaverisetooneofthelargest andmost abrupt economictransformations ofanymajorindustrial countryinthepostwarera. Between November1989andJanuary1991,unificationwent from beingalmost inconceivabletoan establishedfact. Duringthesemonths economicandpolitical observers aroundtheworld scrambledtoassess thelikelyimplications ofunificationforGermanyandtherest oftheworld. Oneapproachtoexploringtheeconomicimplications was throughsimulationof macroeconometricmodels. Unificationposedamajorchallengetomodelers,however,because it promisedafundamental structural changetotheGerman--andparticularlytheEast German-economy. This fundamental changewas ofanaturenot observedbefore. Forthefirst timea socialist plannedeconomywas beingtransformedintoacapitalist market economy,andthe transformationwas occurringalmost overnight. Giventhepotential importanceofunificationfortheworldeconomy,from thefirst there wereanumberofsimulationstudies ofits effects. However,traditional models withbackwardlookingexpectations werenot well suitedtothetask,becausetheydidnot allowforsharp anticipatorymovements inasset prices. Alreadyearlyin1990financial markets weretryingto anticipatethefutureeffects ofunification,beforeanyofits aspects (currencyunion,fiscal transfers,etc.)hadoccurred,andthis producedasharpriseinGermaninterest rates. 1Gagnonis SeniorEconomist intheDivisionofInternational Finance,BoardofGovernors of theFederal ReserveSystem; Massonis Assistant DirectorintheResearchDepartment, International MonetaryFund; andMcKibbinis ProfessorofEconomics at AustralianNational UniversityandNonresident SeniorFellowat theBrookings Institution. Wearegrateful toClaire Adams forresearchassistance,toTessavanderWilligenforsupplyingsomeofthedata,andto WilfriedJahnke,Russell Kincaid,DonoghMcDonald,andEllenMeadeforhelpful comments. This paperreflects theviews oftheauthors andshouldnot beinterpretedas representingthe views oftheBoardofGovernors oftheFederal ReserveSystem,theInternational Monetary Fund,AustralianNational University,theBrookings Institution,orothermembers oftheirstaffs.

-2- Inthis paper,wereport onsimulationstudies preparedearlyin1990usinganew generationofforward-lookingmulti-countrymodels withrational expectations andexplicit attentiontobothstockandflowequilibrium. Threemodels weretheMULTIMODmodel maintainedbyPaul Massonandcolleagues at theInternational MonetaryFund; theMSG2model maintainedbyWarwickMcKibbinat theBrookings Institution; andtheMX3model maintained byJosephGagnonat theBoardofGovernors oftheFederal ReserveSystem. Thesemodeling groups investigatedavarietyofaspects ofGermanunificationwithmodels havingsomebasic similarities but usingverydifferent approaches to incorporatingtheshock ofunification into theirmodels. Thegoal ofthis paperis todeterminewhat lessons canbelearnedbycomparingtheearly model simulations ofunificationwiththemacrodataforGermanyandtherest oftheworldinthe fiveyears sinceunification. Thepotential lessons canbedividedintotwocategories: First,what aspects ofunificationwerewell capturedbythemodels,andcantheassociatedmodel properties beusedtoinform policymakers inthefuture? Second,what aspects ofunificationwerenot adequatelydescribedbythemodels, and can modelers improvetheirfutureperformance? In particular,wherethemodels didnot perform well,was this duetogovernment policies that were different from thoseassumed,or,instead,tostructural deficiencies inthemodels orinthe implementationoftheunificationshocks? II.An Overview of Developments SinceUnification This sectionprovides acapsulesummaryoftheevolutionofthemainmacroeconomic variables sinceeconomicunificationinJuly1990. Forsomevariables,unificationhas produced breaks inseries,whichmakeinterpretationmoredifficult. Nevertheless,anoverall picture emerges ofasharpfall inEast Germanoutput,followedbyaperiodofsustainedgrowththat is fasterthanintheWest. Investment is relativelystrongintheEast,but largefiscal transfers continuefrom West toEast. West Germanyfacedaninitial periodofstrongdemandand

-3inflationarypressures,counteredbyarestrictivemonetarypolicythat raisedinterest rates in Germanyandacross Europe,andcontributedtostrains intheEuropeanMonetarySystem (EMS) andthecrises of1992-93. AEuropeanrecession occurredin1992,followedbyarecoveryof growthwhichnevertheless leaves unemployment highinanumberofcountries,including Germany. A.Output andInflation Unificationwas accompaniedbyacontinuationofthehighoutput growthseenin GermanyandinEuropegenerallyinthelate1980s (Chart 1). However,demandpressures in Germanybegantoput upwardpressureoninflationinGermany,raisingtherateofgrowthofthe CPItoalmost 6percent bytheendof1991. Theboom inoutput was followedbyashort,sharp recessionin1993,witharesumptioningrowthin1994. InEast Germany,therewas aninitial sharpfall inoutput,as theshift tothemarket economymademanyproducts uneconomic. However,startingin1991,annual output growthin theEast has beeninthe5-10percent range,consistentlyhigherthanintheWest (Chart 2). Output intheEast was sustainedbystronginvestment growth(Table1). Inflationrates inthe East werehighinitially,accompanyingmajorchanges inrelativeprices ofvarious goods and services,as well as largewageincreases. By1994,however,therateofinflationwas comparable tothat intheWest. B.MonetaryandExchangeRateDevelopments At thetimeit was announcedinFebruary1990,theconversionrateofoneostmarktoone deutschemark(DM)forassets2andwages generatedconcerns that this wouldcreateinflationas well as makeEast Germansalaries uncompetitive. Theseconcerns weresoonseentobe unfounded,as thesizeoftheGermanmoneysupplyincreasewas not greatlyout oflinewiththe sizeoftheEast Germaneconomy. Moreover,thoughEast Germansalaries werenot competitive 2Theone-to-oneconversionrateappliedtoalimitedamount ofmonetaryclaims only--the averageconversionrateforthebanks'balancesheets was about 1.8to1.

-4at theconversionrate,theyweremadeevenless competitivebythesubsequent actions ofthe tradeunions,whichaimedtoequalizesalaries across Germanyandobtainedlargesubsequent wageincreases (seebelow). Thoughtheconversionofostmarks didnot present aninflationaryproblem,theGerman moneysupplydidgrowstronglyinthe1991-94period,at times exceeding10percent ona twelve-monthbasis (Chart 3),as excess demandledtostronggrowthinnominal income; in addition,therewereshifts out ofotherassets intoM3,makingthelatteraless reliableindicator. SinceGermanmonetarytargets involvedkeepingmonetarygrowthfortheyearintherangeof 3.5-5.5(1992),4.5-6.5(1993),and4-6(1994),theBundesbankraisedshort-term interest rates steadilyoverthe1990-92period(Chart 3,bottom panel). However,long-term rates inGermany starteddecliningfrom theendof1990. Short-term interest rates inotherEuropean countries alsoroseoverthis period,andin particularincountries that participatedin theExchangeRateMechanism (ERM)oftheEMS. Theseincreases ininterest rates wereaccompaniedbyincreasedtensions intheERM,sincethe cyclical situationofcountries outsideGermanydidnot justifythesametight monetarypolicy pursuedbytheBundesbank,leadingtodoubts about theircommitment totheirparities against theDM. TheSeptember1992ERM crisis ledtothewithdrawal oftheU.K.poundandthe Italianlirafrom theERM andaseries ofsubsequent devaluations ofcurrencies that remainedin themechanism,inparticulartheSpanishpeseta,thePortugueseescudo,andtheIrishpound. Furthertensions inthesummerof1993ledtothewideningofERM bands to15percent earlyin August,andseveral currencies depreciatedbeyondtheirprevious lowermargins. As aresult,the DM appreciatedstronglyin1992-93inreal effectiveterms,bothagainst ERM currencies and moregenerally(Chart 4). Already,strongdemandpressures inGermanyhadledtoasharp declineinits current account position,from asurplus ofover4percent ofGDP in1989toa deficit ofaround1percent inthe1991-94period(Chart 4). C.Wages andEmployment

-5- Theboom ineconomicactivityinWest Germanyledtoasteadyreductionofthe unemployment rate,toabout 5.5percent in1991(Chart 5). Excess demandforlaborwas associatedwithsustainedgrowthinreal wages. InEast Germany,theshockofunification producedalargeincreaseinunemployment,from thenegligiblelevels that prevailedbefore unification. Despiteunemployment rates that werepersistentlyinthe10-20percent range,real wages grewverystronglyinEast Germany(Chart 6). From 1991Q1to1994Q4,theaverage annual increaseinreal wages was about 13percent. Thus,initial disparities inunit laborcosts wereexacerbatedbynegotiatedwageincreases,and,as aresult,therewas noemployment growthuntil 1994,evenrelativetothelowlevels ofemployment that prevailedsoonafter unification. D.Fiscal Policy It was clearfrom thestart that unificationwouldinvolvelargegovernment transfers from theWest totheEast. However,theirsizeandpersistenceweregreaterthanmost initial estimates. Total transfers,eithergross ornet offederal taxes paidinEast Germany,grew steadilyoverthe1991-94period,andby1995,theyconstituted5percent ofWest GermanGDP and41percent ofEast GermanGDP (Table2). About 60percent ofthetransfers constituted support forconsumptionandsocial purposes,and40percent was support forpublicandprivate investment. Thelatterincludes investment subsidies andgrants,accelerateddepreciation,and loans forinvestment purposes onsubsidizedterms. Despitetheenormous burdenonthebudget,Germanywas relativelysuccessful in limitingtheincreaseintheoverall budget deficit,throughacombinationofcuts inother spending(includinglowpublic-sectorwageincreases),and,especially,throughtax increases (Table3). Animportant component ofthelatterwas the"solidarity"surchargeonincometax of 7.5percent reimposedin1995. As aresult,thegeneral government deficit inthat yearis estimatedat 3.6percent ofGDP,withtheinclusionforthefirst timeofinterest costs ofthe Treuhand(whichhadbeenoffbudget before).

-6- E.RestructuringtheEast GermanEconomy Despitemajorprogress inintegratingtheEast Germaneconomy,thetasks remainingare considerable. Clearly,afirst prioritywas toreducetheextent ofstateownershipandtoconvert theeconomytoprivateownership. This was thetaskoftheTreuhand,whichhas restructured, sold,and,wherenecessary,closedexistingstate-ownedenterprises. Despiteinitial problems due tothelackoftransparencyconcerningpropertyrights andtheneedtoattempt restitutionof propertyexpropriatedbytheformercommunist regime,it has achievedits mandate,having disposedofover13,000businesses. Moreover,newinvestment has beenflowingintoEast Germany,andtherearereports that newplants operatingintheeasternpart ofGermanyareas profitableas thoseintheWest. However,averageprofitabilitylevels clearlyremainmuchbelowthoseintheWest. Usingdata onpotential output andpotential man-hours-workedfrom theBundesbank(MonthlyReport, August 1995),output perman-hourat normal utilizationrates amountedin1994to64.4DM (at 1991prices)intheWest,andonly21.5DM intheEast. Thus,East Germanlaborproductivityis onlyabout one-thirdthat intheWest. Clearly,anextendedperiodofinvestment is neededto raisethecapital stocksufficientlytoequalizeproductivity(andalso,presumably,toprovide traininginareas not emphasizedbytheformerregime). Thetransitionprocess will alsohaveto involveareductioningovernment subsidies andincometransfers totheEast. Hencethe challenges remainlarge. III.TheModels and Methods of SimulatingtheImpacts of Unification Inthis sectionwesummarizethemodels usedintheearlystudies ofGermanunification andgiveanoverviewoftheapproachandresults ofeachoftheoriginal studies. This overview is important becausethestudies wereeachundertaken independentlyin aperiod ofgreat uncertaintyabout howtheunificationprocess was goingtodevelopbothpoliticallyand economically. Onecommonaspect ofthestudies is that unificationwas viewedas involvinga

-7substantial aggregatedemandshocktoGermany. Thestudies focusedonimportant issues that wouldlikelyemergesuchas thelargeworseningoftheGermancurrent account,highinterest rates inGermanyandrisingreal long-term interest rates throughout theworld,andsignificant pressuretobeplacedoncountries withintheERM as theyattemptedtomaintainparitywhilethe DM appreciatedsignificantly. Thethreemodels discussedinthis papershareanumberofcommonfeatures,which,in turn,generateahighdegreeofsimilarityintheirsimulations ofunification. Oneofthekey characteristics that differentiates thesethreemodels from manyotherempirical macroeconomic models is that eachassumes that expectations are"rational"inthesensethat expectedfuture variables areset equal tothemodel's predictionofthesevariables intheabsenceofshocks. The models focus onthemaincomponents ofthenational incomeaccounts ofindividual countries or groups ofcountries aggregatedtogether. Consumptionis determinedbyexpectedpermanent incomeandthereal long-term interest rate. Investment is drivenbythecurrent andexpected futuremarginal product ofcapital relativetothecost ofcapital,subject toadjustment costs. Government spendingis exogenous,but atax reactionfunctionstabilizes theratioofgovernment debt toGDP. Goods producedindifferent industrial countries areassumedtobeimperfect substitutes,sothat tradeflows respondtorelativeprices andaggregatedemandintheimporting country. Intheshort run,output maydeviatefrom potential. Inthelongrun,prices adjust toreturn output toits sustainablelevel,whichis afunctionofcapital,labor,andtechnology. Price adjustment is primarilydeterminedbyanexpectations-augmentedPhillips curveoraforwardlookingstaggered-contracts model operatingeitheronprices orindirectlythroughwages. Stocks ofphysical capital,government debt,andnet foreign assets arecumulated from therelevant flow variables. Technologyandlaborsupplyareexogenous,at least inthelongrun. Thelong-run equilibrium is essentiallythat ofaSolowgrowthmodel. Long-term interest rates aredetermined

-8byexpectedfutureshort-term rates. Capital is assumedtobeperfectlymobilebetweenindustrial countries sothat exchangerates obeytheopeninterest paritycondition. Theprimaryareas ofdifferences across themodels arethecountrycoverage,thedata frequency,theadjustment dynamics,theestimatedcoefficients,andthespecificationofmonetary policy. A.MULTIMOD MULTIMODincludes separatesubmodels foreachoftheGroupofSevencountries,for theremainingindustrial countries as agroup,andforthedevelopingcountries (dividedinto capital exportingandcapital importingcountries).3 Themodel uses annual data. In MULTIMOD,consumptionbehaviorincorporates bothaforward-lookingpermanent income component andamyopiccomponent that responds tocurrent disposableincome. Thesensitivity ofconsumptiontocurrent incomepermits fiscal policytobeexpansionary,as intraditional Keynesianmodels,thoughtoanextent that depends ontheestimatedcoefficient ofdisposable income. Inmost submodels,themonetaryauthorityis assumedtomoveshort-term interest rates inresponsetodeviations ofthemainmonetaryaggregatefrom its targetedlevel. Inthecaseof Germany,this aggregateis M3. InthecaseofotherERM countries (France,Italy,andthe smallerindustrial countrygroup)themonetaryauthoritymoves theshort-term interest rateto limit deviations intheexchangeratefrom its paritywiththeDM. TheapproachusedinsimulatingGermanunificationinMULTIMODwas totreat the excess ofspendingoveroutput inEast Germanyandthemigrationfrom theEast totheWest as themain"shocks"totheglobal economyassociatedwithunification.4 It was assumedthat,as of 1990,thescaleofthetransfers from West GermanytotheEast was fullyanticipated. Noattempt 3SeeMasson,Symansky,andMeredith(1990)foradescriptionofMULTIMODMarkII. 4SeeMassonandMeredith(1990a,1990b). Thetwostudies differinthefollowingways: the earlier,unpublishedversionusedpreliminaryestimates ofthesizeoftransfers from West toEast Germanythat weresmallerthanthoseinthepublishedpaperbyabout one-quarter,whilethe simulationofa"Europeanmonetarypolicy"was removedfrom thepublishedversion.

-9was madetoexpandMULTIMODtoincludeEast Germany; instead,exogenous inputs were usedtoquantifytheincreaseindemandforWestGermangoods (andthoseofothercountries) that was expectedtoresult from Germanunification. Turningtothemagnitudeoftheshocks,Table4presents twosets ofassumptions ofthe fiscal transfers from theWest andtheincreasednet import demandinEast Germanythat were assumedtoresult from unification.5 Inthefirst one,whichwas termedthe"referencecase," investment was assumedtoproceedat arapidenoughratetoraiseoutput perworkerinEast Germanyto80percent ofthelevel intheWestby2001. Inthesecond--less optimistic--scenario, investment intheEast was assumedtobelowerbut savingwas alsolower,andoutput perworker reachedonly60percent ofWest Germany's level in2001. It was assumedthat intheabsenceof unification,thecurrent account positionofEast Germanywouldhavebeenroughlyinbalance,so thefigures inTable4constituteadditional demands onworldsaving. Thebaselinealsoassumed potential GDP growthof2percent intheEast and2.75percent intheWest. It shouldbenoted, bycomparisonwithTable2,that boththemagnitudeandthepersistenceofthetransfers toEast Germanyweregreatlyunderestimatedinthesimulations. Anothersignificant aspect ofunificationwas there-establishment offreemobility betweenEast andWest Germany,resultinginsubstantial westwardmigration. Thelast few months of1989sawlargepopulationflows from theEast,andsubstantial migrationcontinued earlyin1990. Inthereferencescenario,net migrationfrom East toWest Germanywas assumed tobe280,000in1990,100,000in1991,70,000in1992,40,000in1993,and20,000ayear thereafter. Intheless optimisticscenariowithlowerinvestment intheEast,net migrationwas assumedtobethesamein1990-91,but tobeconsiderablyhigherfrom 1992onward: 270,000in that year,220,000in1993,anddecliningto90,000intheyear2001. 5Theseassumptions werebasedoncontemporaneous workbyMcDonaldandThumann (1990).

-10- Migrationwas expectedtoleadtoincreases inbothaggregatedemandandsupplyinWest Germany; thecorrespondingdeclines intheEast wereembodiedinhigherprojectednet imports. InMULTIMODpotential output is describedbyaproductionfunctionthat depends oncapital andlaborwithconstant returns toscale. Foragivencapital stock,migrationwouldaffect potential output throughtheinducedincreaseinthelaborforce,times themarginal product of labor; inthelongerrun,capital shouldincreasewiththelaborforceandpotential output rise proportionately. Thelaborforceincreasewas calculatedas thepopulationincreasetimes the participationrate. Inthereferencecase,thelevel ofpotential output was projectedtobe 1.25percent higherinWest Germanybytheyear2001thanit wouldhavebeenintheabsenceof migration. Intheless optimisticscenario,it was projectedtobe4percent higher,as aresult of thelargermigration. Thesimulations highlightedthreefactors that wereimportant indetermininghowmuch ofanincreaseofdemandfrom East Germanywouldshowupintheform ofincreasedoutput in West Germany,howmuchas lowercombinedGermannet exports,andhowmuchas higher inflation: (1)thestanceofmonetarypolicy,(2)theinfluenceofthelevel ofcapacityutilization oninflation,and(3)theinterest elasticities ofdomesticcomponents ofdemand. It was recognizedthat theconduct ofmonetarypolicycouldhavebeenaffectedby currencyunionbecause(amongotherreasons)theincomevelocityofmoneymight not bethe sameinthetwoparts ofGermany. Ratherthanattemptingtoquantifythoseeffects,it was assumedforthepurposes ofthesimulations that targets wouldbeappropriatelyadjustedtotake intoaccount velocityshifts andotherfactors that wouldotherwiseaffect therelationshipbetween interest rates andeconomicactivity. Concerningtheeffects ofanincreaseindemandoninflationandoutput,productive capacityis not anabsoluteconstraint onoutput inMULTIMOD. Instead,thehigheris therateof capacityutilization,thegreaterarepressures oninflation. Moreover,themodel implies afairly flat (andlinear)output-inflationtradeoff. Alternativesimulations usingasteeper,nonlinear

-11tradeoffyieldedslightlymoreinflationandslightlyless output thanthereferencescenario,but theoverall natureoftheresults was quitesimilar. Theeffects oftheunificationshockdependimportantlyontheinterest elasticities of savingandinvestment. TheMarkIIversionofMULTIMODhadquitehighelasticities. Some otherevidenceonGermanyandothercountries suggestedthat savingandinvestment might not beas sensitivetointerest rates. TheMULTIMODsimulations wereperformedusingrevised equations forconsumptionandinvestment (subsequentlyincorporatedinthestandardmodel) that embodiedlowerinterest-rateeffects thanintheoriginal MarkIImodel,makingtheresults moreconsistent withthis empirical evidence. 1.TheReferenceScenario(Table5) Thereferencesimulationofunificationassumedthat net imports intoEast Germany wouldincreasebyamounts givenincolumn3ofTable4. This increaseindemandwas assumed toshowupinthefirst instanceinincreasedexports byWest Germany(twothirds oftheamount) andbyothercountries (theremainingonethird, allocatedonthebasis ofhistorical shares in imports ofWest Germany). Thecombinedgovernment deficit as reportedhereincludes all unification-relatedgovernment expenditures,as well as higherinterest payments duebothtoa largerdebt stockandtohigherinterest rates. Inthereferencecase,tax ratesareassumedtobe thesameas inthebaseline,whichassumes nounification. However,tax revenues areelastic, andincreaseroughlyinproportiontoGDP. Thesimulations alsoincludetheprojections of migrationfrom theEast totheWest describedabove,andtheresultingincreases ofpotential output intheWest. 2.ALess OptimisticScenariowithSlowerGrowthinEast Germany(Table5) Inanalternative,less optimisticscenarioforunification,investment was assumedtobe less buoyant. As aresult,productivitygrowthconvergedless quickly,andbytheyear2001,the productivitygapbetweenEast andWest Germanywouldstill beabout 40percent. Net imports bytheEast werenot verydifferent initiallyfrom thoseinthereferencecase(seeTable4),but the

-12- East's tradedeficit wouldpersist longerbecauseoutput wouldnot riseas muchinthemedium term. Correspondingly,incomeandsavingarealsolowerthere. Inthis scenario,real incomes areassumedtobelowerinEast Germany,andemigration higherthaninthereferencescenario: extranet emigrationfrom theEast totheWest amounts to 200,000in1992,180,000in1993,andgraduallydecliningamounts thereafter(inadditiontothe projectedmigrationinthereferencescenario). Government expenditures intheWest are assumedtobehigheras aresult oftheincreaseinpopulationrelativetothereferencescenario (duetoincreasedexpenditureonhousingandsocial services,forinstance). Unemployment benefit payments arealsohigherintheEast,as aresult ofhigherunemployment. 3.AnERM Realignment (Table5) Thereferencescenariosuggestedthat an appreciationoftheDM ofabout 4percent against theU.S.dollarmight result from unification. Withfixedcentral parities withrespect to othercurrencies participatingintheERM,real appreciationoftheDM resultedfrom a combinationofnominal appreciationagainst non-ERM currencies (principallytheU.S.dollar andtheyen),increases inprices inGermany,andatendencytodeflationinotherERM countries. It was suggestedinMassonandMeredith(1990b)that thetighteningofmonetaryconditions in otherERM countries might beavoidedbyarealignment vis-à-vis theDM,alsoinprinciple permittingasmootherallocationoftheincreaseddemandfrom East GermanyamongEuropean countries. Theupwardrealignment oftheDM wouldtendtoremovesomeoftheshort-run pressureonexistingcapacityinGermany. Thesefavorableeffects,it was argued,would howeverhavetobebalancedagainst thenegativeeffects ofhigherinflationintheshort runin otherERM countries,andconsequentlyapossibleloss ofcredibilityoftheircommitments to pricestabilityandto"hardcurrency"policies. Arealignment scenariowas presentedinwhicha depreciationof4percent oftheotherERM currencies vis-à-vis theDM occurredin1991,but hadunfavorableeffects onexpectations offutureexchangeratemovements andinflation differentials. Specifically,theotherERM currencies wereexpectedtodepreciatebyafurther1

-13- 1/2percent ayearagainst theDM intheyears followingtheinitial realignment,similartothe periodicrealignments that wereobservedintheearlyyears oftheEMS. B.MSG2 TheversionoftheMSG2model usedintheoriginal studyincludedsubmodels ofthe UnitedStates,Japan,Germany,theRest oftheEMS (REMS),theRest oftheOECD,thenon-oil developingcountries,andtheoil-exportingdevelopingcountries usingannual data.6 As in MULTIMOD,consumptioninMSG2depends onbothpermanent incomeandcurrent disposable income. Inmost submodels,themonetaryauthoritysets anexogenous pathofthemainmonetary aggregateandtheshort-term interest rateis determinedthroughthemoneydemandequation. In theREMS submodel themonetaryauthoritypegs theexchangeratetoits paritywiththeDM. Thegoal ofthestudyreportedinMcKibbin(1990)andupdatedinMcKibbin(1992)was toexaminetheimpact ontheglobal economy,in particularglobal asset markets,ofthereal transferrequiredtoraiseEast Germanlivingstandards towardthoseofWest Germany. In particularthepaperfocusedonthedifferencebetweenfinancingthereal transfers byfiscal versus monetarymeasures inGermanyandtheimpact oftheERM ontheadjustment oftheworld economy. Thepaperattemptedtoplacesomepreliminaryempirical magnitudes onpossible alternativescenarios forGermanunification. Thepaperfocusedontheimplications for aggregatedemandandsupplyinaunifiedGermanyandhowthesespill overintotherest ofthe world. It was not intendedtoprovideaspecificforecast ofthelikelyoutcomes ofunificationbut tofocus onalternativescenarios. Afurtherobjectivewas todirectlyaddress thedebateinthe financial press about "unexplainable"movements ofreal long-term interest rates andexchange rates from theendof1989. Thepaperillustratedthat theco-movements ofasset prices within countries andbetweencountries dependonthenatureofactual andexpectedshocks. Thebasichypothesis ofthepaperwas that thepolicyresponseofmakingtheEast Germancurrencyconvertiblewoulddolittletostem thetideofpeoplemovingfrom East to 6For a description of MSG2, see McKibbin and Sachs (1991).

-14- West. What was inevitablewas goingtobeatransferofphysical capital from West Germanyto East Germanyinordertoraisethemarginal product oflaborinEast Germany. It was arguedthat this transfercouldbeaccomplishedinavarietyofways. Firstly,attractedbylowerEast German laborcosts,privatecapital couldflowintoEast Germany. Secondly,ifwages inEast Germany wereraisedartificiallytoWest Germanlevels,theWest Germangovernment couldeffect the transferofreal resources from West toEast viadirect subsidies toemployment intheEast sothat privatecapital wouldfindit attractivetoinvest theredespitethelowerlaborproductivity. Under eitherscenario,thereweregoingtobesignificant fiscal implications ofgrantingEast Germans West-Germanunemployment andsocial securitybenefits. Tounderstandthedifferent implications foraggregatedemandandsupplythepaper examinedanumberofscenarios inturn. 1.Fiscal Expansion(Table6) Withelections loominginWest Germanyandpromises oftax cuts,it was arguedthat thesefiscal measures werelikelytobefinancedthroughalargerGermanfiscal deficit. Thefirst scenarioconsideredtheconsequences ofasubstantial expectedGermanfiscal expansion resultingfrom direct subsidies orunemployment compensationpaidtoEast Germany. Thefiscal expansionwas assumedtoequal 1.7percent ofbaselineGDP in1990and3.3percent ofbaseline GDP from 1991onwards. Thesescenarios focused ontheimplications ofexpansionaryfiscal policyfortheWest Germaneconomyandtheymodeledthetransferas anincreaseinWest Germangovernment consumptionthat was handedovertoEast Germans. 2.Fiscal ExpansionwithERM Realignment (Table6) Thesecondscenariowas thesameas scenario1except that theDM was allowedto appreciateagainst theU.S.dollarandagainst theotherERM currencies whichremain approximatelyunchangedrelativetotheU.S.dollar. 3.MonetaryExpansion(Table6)

-15- Thethirdsimulationwas arelaxationofGermanmonetarypolicy. This couldbe interpretedas theresult ofcurrencyunificationwithanunrealisticallystrongexchangeratefor East Germanprices andwages andnootherdirect measures forEast Germany. Inthis case,it is likelythat asevereslowdownwouldresult inEast Germany(still not explicitlypart ofthe model)andthat this wouldleadtopressureontheBundesbanktoeasepolicy. Abstractingfrom moneydemandchanges originatingintheEast,scenario3consideredanincreaseinthemoney growthrateof2percent peryear. 4.SupplySideEffects (Table6) Thefinal scenariowas intendedtocapturethepossiblemacroeconomicconsequences for theworldeconomyofthesupply-sideeffect ofacompletelyunified Germany; boththeEast Germancapital stockandworkforcewereincorporatedintotheWest Germaneconomy. This simulationassumednoadjustment infiscal andmonetarypolicies inthenewGermanyinexcess ofthepolicies inplaceinWest Germanybeforeunification. Inotherwords,thefiscal balance changes littleandthemoneysupplywas raisedbyenoughtokeepGermaninterest rates almost unchangedinthefaceoftheincreaseinaggregatesupply. It assumedthat East Germanlabor was integratedintoWest Germanyat theWest Germanwageratedespitealargeproductivity differential. Thestickiness ofwages initiallyledtosevereunemployment,despiteassumingfull employment ofexistingEast Germanphysical capital. Inthelongrun,wages adjustedand physical capital was put inplacetoabsorbtheunemployedaccordingtostandardneoclassical growththeory. Intheshort run,severedislocationofresources resulted. This scenarioshowed theextent oftheproblem ofattemptingtointegratetwoeconomies withverydifferent productivitylevels,quicklyandwithout restrictions. It was argued that theresults showedclearly that attemptingfull integrationinthewaypredictedwouldbepotentiallydisruptive,but it would not automaticallyberuledout duetopolitical considerations. It was arguedtobeauseful benchmarkforevaluatingalternatives. C.MX3

-16- TheMX3model includes submodels oftheUnitedStates,Germany,Japan,andtherest oftheworld(ROW)usingquarterlydata.7 UnlikeMULTIMODandMSG2,MX3does not includeacomponent ofconsumptionthat is proportional tocurrent disposableincome. Instead, consumptioninMX3follows alaggedadjustment toits forward-lookingpermanent incomelevel This propertymakes output inMX3muchless sensitivetofiscal policythanintheothertwo models. InMX3,themonetaryauthorities areassumedtomoveshort-term interest rates in responsetodeviations ofatarget variablefromits targetedlevel. Inthesimulations ofGerman unification,monetarypolicywas assumedtotarget thepricelevel inmost cases. InMX3,Germanunificationwas modeledbyaugmentingthesupplyequations inthe West Germanmodel toincludeEast Germanfactors ofproduction. This modificationrequired not onlychangingthestructureofthesupplysidebut alsochoosingthebaselinelevels of important variables andparameters. Choices regardingthesemagnitudes determineboththe speedwithwhichtheEast Germaneconomyconverges withtheWest Germaneconomyandthe strengthofthespillovereffects onothercountries. Thecritical assumptionwas that East Germanybrings relativelymorelaborthancapital totheunion,creatingasubstantial differential betweenthecapital-laborratios inthetworegions. As aresult,unificationincreasedGerman aggregatedemand--primarilythroughincreasedinvestment andconsumption--morethanit increasedGermanaggregatesupply. Twostudies ofGermanunificationwereconductedwithMX3. Theresults presented herearedrawnfrom Adams,Alexander,andGagnon(1993),whichis thepublishedversionof International FinanceDiscussionPaperNo.421,January1991. Anearlierstudy,Alexanderand Gagnon(1990),reachedsimilarconclusions but did not considertheimplications ofunification forERM countries.8 Onedifficultyinexaminingtheeffect ofunificationonERM countries is 7For a description of the MX3 model, see Gagnon (1991). 8Othersignificant differences betweenthetwostudies arethat thelaterstudyassumedlower values forEast Germancapital andtechnologyandit allowedforpersistent unemployment in East Germanyas unions pressedtocatchuptoWest Germanwagelevels. Thesedifferences led

-17that MX3consolidates thenon-GermanERM countries withall othercountries intheROW block. Theapproachtakenwas toobtainsomeboundaries ontheERM effect byrunning simulations inwhichROW has afreelyfloatingexchangerateandcomparethem tosimulations inwhichROW pegs its exchangeratetoGermany's. Themost significant changes tothemodel involvedaddingseparateexpressions forEast Germancapacity,fixedinvestment,andlaborsupply. ProductivecapacityinEast Germanywas modeledas aCobb-Douglas functionofEast Germantechnology,capital,andlabor. TheEast Germanlaborforcewas set at 30percent oftheWest Germanlaborforcein1990andit was assumedtoremainconstant overtimewhiletheWest Germanlaborforcewas assumedtogrow at 0.5percent peryear. East Germanlaborproductivitywas assumedtobeone-thirdthat ofWest Germany. This lowerproductivitywas modeledas duetobothlowertechnologyandalower capital-laborratioinEast Germany. Afterunification,theEast Germantechnologyfactorwas assumedtoexogenouslyconvergetotheWest Germanlevel overfiveyears. Alternative simulations wereconductedtoexaminethesensitivityoftheresults todifferent assumptions about thelevel andrateofconvergenceofEast Germantechnology. Investment inEast Germany was modeledas astock-adjustment process designedtograduallyraisetheEast Germancapital stockuntil themarginal product ofcapital is equal tothat ofWest Germany. IntheWest GermansectorofMX3,equilibrium laborsupplyis simplyaconstant fraction oftheexogenous laborforce. Inordertomodel theinitial increaseinEast German unemployment,an"effective"East Germanlaborsupplywas constructed that was anegative functionoftheexcess ofWest GermanoverEast Germanwages. Labordemandinbothparts of Germanywas determinedbyaggregatedemandintheshort runandbyequalizingthemarginal product oflaborwiththereal wageinthelongrun. Together,labordemandandlaborsupply impliedthat theNAIRUwouldbehigherinEast Germanyas longas wages werelowerthanin toamodest quantitative--but noqualitative--changeinthemodel simulations ofkeyGerman macroeconomicvariables.

-18- West Germany. Therewas assumedtobenonet migrationbetweenthetwoparts ofGermany after1990.9 Thesimulationresults forunifiedGermanyareexpressedrelativetoabaselineinwhich thelaborproductivityofEast Germanyis fixedat one-thirdthat ofWest Germany. The unificationshockis implementedbyaugmentingtheGermansupplysectoras describedabove andbyraisinggovernment spendingexogenously. Theseshocks haveanimmediateeffect on privateaggregatedemand. Consumptionincreases becausethenewexpectationofamore productivefutureEast Germanyraises permanent income. Investment increases duetothe highermarginal product ofcapital inEast Germany. 1.TheReferenceScenario(Table7) Inthereferencescenario,fiscal policywas modeledbysettinggovernment spendingin unifiedGermanyequal tothat ofbaselineWest Germanyonaper-workerbasis. This implieda largeincreaseingovernment spendingrelativetothecombinedbaselines ofEast andWest Germany. Overtime,thetax reactionfunctionensures that theratioofgovernment debt toGDP returns toits baselinelevel. Insimulations,thetax reactionfunctioneliminatedthefiscal deficit morequicklythaninthereal world,sincealargedeficit still remains in1996. Alternative simulations wereconductedwithamorepersistent fiscal deficit,but themainconclusions were littleaffectedbecausetheforward-lookingconsumers inMX3areonlymildlysensitivetothe timepathoftaxes,i.e.MX3is closetoRicardianequivalence. Inthereferencescenario, exchangerates arefreelyfloatingbetweenall countries. 2.ERM Scenario(Table7) Scenario2augmentedscenario1withanapproximationoftheERM effect. Inthis scenario,theshort-term interest rateinROW was assumedtotarget theexchangeratewith Germany. 9An alternative simulation showed that the results for unified Germany were relatively insensitive to migration (1 million workers were assumed to migrate from East to West over a five-year period).

-19- 3.ERM ScenariowithAccommodativeMonetaryPolicy(Table7) Scenario3combines scenarios 1and2withaccommodativeGermanmonetarypolicy. In this scenariotheBundesbankraises its pricelevel target by1percentagepoint peryearforthree years. IV.Results fromtheModels and KeyIssues Raised in theEarlyStudies InsectionIIIweoutlinedthreestudies,focusingonthecharacteristics ofthemodels,the approachtomodelingunification,andtheissues that wereaddressed. Inthis sectionwepresent somekeyresults from eachmodel anddrawout someoverall lessons from thestudies andtheir implications formodel development andpolicyformulation. Theresults from eachstudy dependedonthemodel usedas well as theassumptions about likelydevelopments inGermany from 1990forward. Asurprisingaspect oftheresults was that therewas agreat deal of consensus across thesestudies about someaspects ofunificationthat subsequentlyemergedas important issues afterthestudies werecompleted. Theseaspects wereignoredbymany commentators at thetimeandarestill ignoredbysomeanalysts whoarguethat noonecould haveforeseentheproblems that emergedinunifyingthetwoGermanies. Tables 5through7present results foranumberofvariables foreachmodel from 1990to 1997orlater. Detaileddescriptions ofeachscenariocanbefoundinsectionIIIabove. Table5 contains standardizedresults forMULTIMOD,Table6contains standardizedresults forMSG2, andTable7contains standardizedresults forMX3. Wehaveselectedvariables that eachmodel canproduce,althoughinsomecases theselectedvariables werenot presentedintheoriginal studies. Inthesecases,theselectedvariables werecalculatedusingothervariables that were includedintheoriginal studies. Thevariables presentedarefor: UnitedGermany(GDP,current account,andfiscal balance),West Germany(GDP,inflation,real effectiveexchangerate,and real long-term interest rate),andotherERM countries (GDP andinflation).

-20- Eachvariableis expressedeitheras apercent deviationfrom baseline(e.g.GDP),a percent-of-GDP deviationfrom baseline(e.g. current account,fiscal balance),orapercentage point deviationfrom baseline(e.g.inflation,interest rates). Eachscenarioproducedbythe models is givenascenarionumberbasedontheassumptions discussedinsectionIIIabove. The scenarioorderingis basedontheoriginal studies andis not necessarilythesameacross models. Thesetables canbecomparedbyfocusingeitheronthedifferences betweenresults oron similarities. However,atruecomparativeanalysis is difficult becauseeachstudyundertook different simulations underdifferent assumptions about policyresponses. Nonetheless,wecan drawout themajormechanisms from eachmodel. First,eachstudypresumedthat therewouldbeasignificant fiscal expansioninGermany althoughthesizeandpersistenceofthefiscal stimulus variedacross models. This aspect of unificationis illustratedforMULTIMODandMX3ineachscenario,andinMSG2inscenarios 1 and2. IntheMULTIMODresults forscenario1and3andall theMX3results thefiscal deterioration lasts forup to 5 years whereas, in theMSG2 scenarios and MULTIMODscenario 2, thefiscal stimulus is morepermanent. Despitethesedifferences inassumptions,theadjustment process is essentiallysimilaracross themodels. Theannouncement ofthefiscal stimulus in Germanyraises real andnominal long-term andshort-term interest rates inGermany. Inroughly comparablescenarios real long-term interest rates risebetween80and130basis points ineach model. Theriseininterest rates leads toacapital inflowthat appreciates theDM inbothreal and nominal terms. Theriseinthereal effectiveexchangerateis somewhat moredispersedacross models,rangingfrom alowof2percent fortheMULTIMODreferencescenariotoahighof11 percent fortheMX3 scenarios that pegtheGerman-ROW exchangerate. Thecapital inflowand exchangerateadjustment arereflectedinadeteriorationintheGermancurrent account of between2and3percent ofGDP across themodels. Thefirst clearmessagefrom themodels was

-21that theadjustment process ofgettingtherequiredresources intoGermanywouldbeviaa deteriorationinthecurrent account facilitatedbyanappreciationoftheDM. Thefiscal aspect ofunificationwouldtendtoraiseGDP andinflationinGermany relativetowhat theyotherwisewouldhavebeen. Incontrast toMULTIMODandMX3,the results forMSG2haveoutput lowerthanotherwiseandinflationlowerin1990but ifcombined withsomemonetarystimulus this result forMSG2disappears. Thereasonforthis result in MSG2from thepurelyfiscal shockis that theanticipatedfiscal program crowds out private investment in1990beforethefull spendingincreasebegins. Inadditiontheappreciationofthe DM lowers importedgoods prices whichreduces inflationinitially. Themodels alsoshowthat theextent ofoutput,price,andexchangeratechangedepends importantlyontheassumptionoftheresponseofGermanmonetarypolicy. BoththeMSG2and MX3models showedthat moremonetaryaccommodationinGermanyafter1990leads toa largerriseinGDP andinflationandless appreciationoftheDM. Anotherlessonthat is commontoall threestudies is theimplicationoftheunification process inGermanyfortheothercountries intheERM. Withupwardpressures ontheDM other ERM countries wereforcedtoraiseinterest rates tomaintainexchangerateparities. This tighteningofmonetarypolicyintherest ofEuropewas contractionaryforGDP andplacedstrain ontheERM. Comparisonofscenario3and1forMULTIMOD,scenarios 2and1forMSG2and scenarios 1and2forMX3clearlyillustratethis point. Theextent ofoutput loss fortherest of theERM differs across models from between0.4%to3.9%(at thepeak)relativetoarangeof0 to0.5%ifanERM realignment was implemented. Again,thesizeofresults varies across the models becauseofmodel differences anddifferent assumptions about theextent offiscal adjustment inGermany. Nonetheless,this important lessonis amplydemonstratedinthese results. Germanunificationwas likelytobecontractionaryfortherest ofEuropeifERM parities weremaintainedandtherewouldbestress withintheERM as aresult.

-22- Theunanimityofthemodel results withrespect totheERM countries reflects four commonproperties sharedbythemodels andtheirimplementationofunification. First, unificationledtoapositiveaggregatedemandshockinGermany. Second,aggregatedemandin eachcountryis biasedtowarddomesticgoods. Third,goods producedindifferent countries are imperfect substitutes. Fourth,Germanmonetarypolicywas assumedtotarget domesticnominal variables without regardforthebehaviorofthesevariables inothercountries.10 Underthese conditions,theERM countries wereforcedtoaccept thesamehighinterest rates andexchange rates as Germanywithout thebenefit ofthefull aggregatedemandshockthat tookplacein Germany. Theimmediateconclusionis that theERM countries facedareductioningrowthand inflation. Oneinterestingareaofdissimilarityacross thesemodels is thelong-runsupplyeffect of unificationonGermany's real effectiveexchangerate. Inthestandardtradeequations usedby most models,exports dependonforeignincomeandimports dependondomesticincome. Since Germanunificationimplies along-runincreaseinGermanincomeas technologytransferand capital accumulationraiseEast Germany's productivitytowardtheWest Germanlevel,unified Germanytends toexperienceanincreaseinimports relativetoexports unless theexchangerate depreciates inreal terms.11 Asupply-induceddepreciationshows upstronglyinMSG2scenario4. Scenarios 1and2 demonstratethat thereal depreciationmaybeoffset byfiscal expansion,but theappreciation 10A scenario reported in Masson and Meredith (1990a) showed that targeting a European monetary aggregate produced smaller negative output effects for the other ERM countries. 11Wyplosz (1991) made this argument in the context of a theoretical model. He also pointed out that a reduction in the German current account balance would lower net foreign assets in the long run relative to their position in the absence of unification, and that this should lead to a further depreciation of the real exchange rate. This net foreign asset mechanism is incorporated into all three sets of model simulations described here.

-23causedbyfiscal expansionis reversedovertimewhilethedepreciationcausedbyhighersupply continues togrow,sothat theDM must depreciateinthelongrun. TheMULTIMODscenarios incorporateaspects ofbothfiscal expansionandsupplyincrease,andtheyexhibit areal depreciationoftheDM,albeit asmall one,in2001. This propertyis absent from theMX3 simulations becausetheincomeelasticityofimports is afunctionofthecyclical positionof income(i.e.,relativetopotential income). MX3assumes that all tradeflows growinproportion toworldcapacityinthelongrun. V.Conclusions and Lessons forPolicymakers and Modelers Table8gives somedataforthemacroeconomicoutcomes overthe5years ofunification, 1990-94,andfortheprevious year. Acomparisonof1989withthesubsequent years gives some indicationoftheresults ofunification--thoughofcourseothershocks ordynamicadjustment couldexplainwhy1990-94differedfrom 1989.12 Thereis somejustificationfordoingso, nevertheless; 1989was anormal yearforGermany,andhenceagoodstartingpoint--thefiscal positionwas inbalance,inflationwas roughlyconsistent withtheBundesbank's long-runtarget, andgrowthwas reasonablystrong. Table8suggests that output growthandinflationinGermanywerestimulatedby unification,that interest rates roseandtheDM appreciated,andthat as aresult thefiscal position deterioratedsharplyandthecurrent account went from asurplus ofalmost 5percent toadeficit ofonepercent ofGDP. Thesearestylizedfacts that accordwell withthemodel simulations reportedabove. As expected,theseeffects seem todieout overtime,thoughinseveral respects theoutcomes,whencomparedto1989,differfrom themodel simulations ofunificationwhen comparedtobaseline. Thesharpest differenceis that output actuallydeclinedin1993. Noneof 12Analternativeproceduretoevaluatethesimulations wouldbetorecreatethebaselines actuallyusedinthesimulations reportedinsectionIII. However,this has thedisadvantageof requiringthreedifferent evaluations; moreover,sincethemodels consideredare,formost purposes,closetolinear,thebaselineusedis irrelevant.

-24themodels predictedarecessionin1993,althoughMULTIMODdidpredict amodest reduction inthegrowthrateofWest Germanyrelativetobaseline. Otherdifferences arepredominantlya matteroftiming,includingthefact that thereal effectiveexchangerateofGermanycontinues to appreciatethrough1994andinflationinGermanypeaks in1992-93. Onepotential explanationforthesedifferences is that noneofthesimulations adequately capturedtheinteractionofinflationandmonetarypolicyinGermany. Theyear-over-year Germaninflationrateincreasedbyover1percentagepoint between1990and1992,13prompting atighteningofmonetarypolicythat ledtoarecessionin1993andthesubsequent declineof inflationin1994and1995. Incontrast,manyofthemodel simulations showedverylittle increaseininflation,andnoneofthesimulations that didhavehigherinflationwereableto matchthetimingoftherealizeddata. It is not clearwhetherthesediscrepancies areduetoerrors intheprice-adjustment sectors ofthemodels ortomisspecificationofthebehaviorand credibilityofthemonetaryauthorities. Anotherimportant questionis theextent towhichtherecessioninEuropecanbe attributedtoGermanunificationratherthantoothercauses. Themodel simulations highlighted thenegativetransmissioneffects onoutput inotherEuropeancountries,andtheactual outcomes indicateaslowdowningrowthfortherest oftheEuropeanUnionwhichoccurs muchearlier thaninGermany. Relativeto1989,thedeclineinoutput growthis 2.8percent alreadyin1991. Inthis instance,therealizeddataappearquitefavorableforthemodel predictions. Ofcourse, otherfactors mayhavebeenat work,includingdrops inconsumerconfidenceacross Europe. Arelatedissueis whetherthemodel simulations clearlyanticipatethetensions inthe ERM,whichultimatelyledtothecrises of1992-93. Thoughthedirections oftheeffects--the increaseininterest rates inGermanyandthenecessityforotherERM countries toraiserates in thefaceofdownwardpressures onoutput--arecapturedinthesesimulations,it must be 13The 12-month inflation rate peaked considerably higher, as discussed above.

-25recognizedthat themodels cannot fullycapturetheinducedeffects onthecredibilityofthese countries'commitments totheircentral parities,northereluctanceofGermanytodefendthose rates. Therefore,themodel simulations canbetakentobeindicativeofpressures,but theydo not predict theresponseofpolicymakers tothesepressures. Afurtherinterestingquestionis whetherthemodel results (ortheexogenous assumptions)correctlycapturedthenatureofthetransformationprocess inEast Germany. There areseveral aspects tothis question. First,it seems that thesizeofthetransfers that wereneeded from theWest was largerthanmost commentators initiallyexpected. Second,wagegrowthin theEast muchexceededwhat was justifiedinterms ofrelativeproductivitylevels. Third,and relatedtothesecond,output declines inEastGermanywerelargerthananticipated. Themodels differintheextent towhichtheyattemptedtomodel developments inEast Germany. TheMULTIMODsimulations usedthenet imports generatedfrom aseparatereal aggregatedemand-supplymodel ofEast Germanyas anestimateoftherelevant shocks tothe West Germanyeconomy. MSG2simulateddifferent aspects ofunificationindependently; oneof theseaspects was thesupply-sideeffect ofaddingEast Germanlaborandcapital toWest Germanyat West Germanwagerates. MX3focusedonthesupply-sideimplications of unificationwithverylittleattentiontothefiscal aspects. Despitethesedifferences,all theinputs tothemodel simulations embodiedan overestimateoftheoutput level inEast Germanyandanunderestimateofthepersistenceof fiscal transfers totheEast. Thesefailures stem largelyfrom anoverlyoptimisticassumption about theeffectivevalueoftheEast Germancapital stock. This mistakewas sharedbymost observers at thetime,andis nodoubt partlyrelatedtotheassumptionthat wages wouldremain low. At higherwagerates,otherwiseprofitableplants weresimplyclosed. Noneofthemodels correctlyanticipatedthesizeandpersistenceofexcess unemployment inEast Germany. MSG2scenario4highlightedtheimportanceofthe unemployment problem,but overestimatedtheadjustment speedtoequilibrium. Thesecond

-26- MX3studymadesomeprogress onthepersistenceofunemployment,but onlybymodifyingthe model structureinlight ofdatafrom late1990. As formigrationfrom East toWest--anissuehighlightedonlyintheMULTIMOD simulations--this ceasedtobeanimportant factorafter1992,as canbeseenfrom Table8. Thus, theassumptionoftheless optimisticscenariothat therewouldbecontinuinglargenet migration was not borneout. However,thereferencecaseinTable5underestimatedmigrationin1990-91 byabout 75,000eachyear. Altogether,themodel simulations providedareasonablyaccurateguidetothedirection andmagnitudeoftheeffects ofunificationonmost keymacroeconomicvariables. However,the model simulations werefrequentlywrongonthetimingoftheseeffects. Themodels tendedto predict thestrongest effects withinayearorsoafterunification,whereas thedatashowthat many oftheseeffects requiredtwoorthreeyears toreachtheirpeak(thoughnot long-term interest rates). This discrepancyis most likelyduetotheimplementationofrational expectations as perfect foresight,inwhichagents areassumedtofullyunderstandandanticipateall aspects of unificationat thebeginningofthesimulations. Onelessonforrational expectations modelers maybetoincorporatelearningbehaviorintothemodel whenanalyzingtheeffect oflargeand unusual shocks suchas Germanunification(fora discussionofhowtodoso,seeHall (1993)). It is always easierwithhindsight toexplainmajoreconomicevents. Inthis paperwe focusedonamajoreconomicevent that is not onlyafascinatingstudyineconomicadjustment in its ownright,but that alsoproduceddataforevaluatingarangeofempirical issues. Inparticular, Germanunificationprovidedus withtheopportunitytoevaluatetheusefulness ofanew generationofmulti-countrymodels that evolvedduringthe1980s. Theexperiment is interesting becausetheconsequences ofthis majoreconomicevent wereanalyzedindependentlybythree modellinggroups usingthesenewmodellingtools andtheresults weremadepublicinthevery earlystages oftheunificationprocess. Theinsights from thesethreemodels arestill relevant in understandingtheactual outcomes ofGermanunificationas well arangeofotherissues currently

-27facingtheworldeconomy. Despitethewidespreaddistrust oflargescalemodelling(primarily becauseofthepoorperformanceofanearliergenerationofdemand-sidemodels inthefaceof theoil priceshocks ofthe1970s)this particularexerciseillustrates that thenewgenerationof thesemodels has provenuseful ingaininginsights ofimportanceforpolicymakers. Ourstudy alsoshows that thewidelyarguedpropositionthat noonecouldhaveforeseentheconsequences ofGermanunificationis anexaggeration. Indeed,therewas agreat deal ofpredictivematerial availablefrom thesethreestudies,as well as otheracademicpapers basedonthesameunderlying theories.

-28- References Adams,Gwyn,Lewis Alexander,andJosephGagnon(1993)"GermanUnificationandthe EuropeanMonetarySystem: AQuantitativeAnalysis,"Journal of PolicyModeling15, 353-392. Alexander,Lewis S.,andJosephE.Gagnon(1990)"TheGlobal EconomicImplications of GermanUnification,"International FinanceDiscussionPapers No.379,Boardof Governors oftheFederal ReserveSystem,April. Gagnon,JosephE.(1991)"AForward-LookingMulticountryModel forPolicyAnalysis: MX3," EconomicandFinancial Computing1,311-361. Hall,Stephen(1993),"ModellingStructural ChangeusingtheKalmanFilter,"Economics of Planning,16,1-13. Masson,Paul R.,andGuyMeredith(1990a)"DomesticandInternational Macroeconomic Consequences ofGermanUnification,"IMFWorkingPaperWP/90/85,September. _____,and_____(1990b)"DomesticandInternational MacroeconomicConsequences of GermanUnification,"ChapterVIofGermanUnification:EconomicIssues,Leslie Lipschitz andDonoghMcDonald,eds.,IMFOccasional PaperNo.75,December. Masson,Paul R.,StevenSymansky,andGuyMeredith(1990)MULTIMODMarkII: ARevised andExtendedModel,IMFOccasional PaperNo.71,July. McDonald,Donogh,andGüntherThumann(1990)"East Germany: TheNew Wirtschftswunder?,"ChapterVofGermanUnification:EconomicIssues,Leslie Lipschitz andDonoghMcDonald,eds.,IMFOccasional PaperNo.75,December. McKibbin,WarwickJ.(1990)"SomeGlobal MacroeconomicImplications ofGerman Unification,"Brookings DiscussionPaperinInternational Economics No.81,The Brookings Institution,May. _____(1992)"TheNewEuropeandits EconomicImplications fortheWorldEconomy," EconomicandFinancial Computing2,123-149. McKibbin,WarwickJ.,andJeffreyD.Sachs (1991)Global Linkages:Macroeconomic InterdependenceandCo-operationintheWorldEconomy,Washington: TheBrookings Institution. Wyplosz,Charles (1991)"OntheReal ExchangeRateEffect ofGermanUnification," Weltwirtschaftliches Archiv127,1-17.

-29- Table1. East Germany: CompositionofGDP (Annual growthrates,inpercent) ____________________________________________________________________________ 1992 1993 1994 ____________________________________________________________________________ Gross domesticproduct 7.8 5.8 9.2 Domesticdemand 15.0 5.4 7.7 Privateconsumption 9.6 2.7 4.5 Government consumption 6.8 -1.2 1.8 Investment 27.8 14.1 16.5 Foreigntradebalance1 -189.9 -199.2 -210.8 Exports 10.2 5.5 22.6 Imports 21.3 5.0 9.4 ____________________________________________________________________________ Source: Statistisches Bundesamt. 1.InDM billion.

-30- Table2. Transfers toEast Germany (InDM billion) ____________________________________________________________________________ 1991 1992 1993 1994 1995 ____________________________________________________________________________ Gross transfers Federal Government 75 88 114 128 151 WesternLaender 5 5 10 14 14 GermanUnityFund 31 24 15 5 -- EuropeanUnion 4 5 5 6 7 Fed.LaborOffice 24 25 15 14 14 PensionInsurance -- 5 9 14 14 Total 139 152 168 181 200 Receipts Taxes andfees 33 37 39 42 45 Net transfers 106 115 129 139 155 Net transfers as apercentageof: West GermanGDP 4.0 4.1 4.5 4.7 5.0 East GermanGDP 51.4 43.8 42.3 40.5 40.6 ____________________________________________________________________________ Source: DeutscheBundesbank, MonthlyReport,July1995.

-31- Table3. Germany: General Government Finances (Inpercent ofGDP) ____________________________________________________________________________ 1990 1991 1992 1993 1994 1995 ____________________________________________________________________________ Expenditure 42.3 48.9 49.6 50.5 50.1 50.6 Revenue 40.4 45.6 46.7 47.2 47.6 47.0 Fiscal balance -1.9 -3.3 -2.9 -3.3 -2.5 -3.6 ____________________________________________________________________________ Source: IMF, WorldEconomicOutlookdatabase.

-32- Table4. MULTIMODSimulations ofGermanUnification: Assumptions ConcerningtheFiscal Cost and IncreasedNet Imports ofEast Germany (Inbillions of1990DM; percent oftrendWest GermanGDP inparentheses) ____________________________________________________________________________ Fiscal Transfers From IncreasedNet Imports of West Germany East Germany Reference Less optimistic Reference Less optimistic case scenario case scenario (1) (2) (3) (4) ____________________________________________________________________________ 1990 95 (3.9) 95(3.9) 116(4.7) 116(4.7) 1991 78 (3.1) 94(3.7) 127(5.0) 122(4.8) 1992-74 42 (1.6) 71(2.6) 101(3.7) 103(3.8) 1995-97 23 (0.8) 53(1.8) 71(2.5) 81(2.8) 1998-2000 -3(-0.1) 41(1.3) 31(1.0) 59(1.9) 2001 -19(-0.6) 33(1.0) 5(0.2) 43(1.3) ____________________________________________________________________________ Source: McDonaldandThumann(1990),Tables 3and6,andcalculations basedon World EconomicOutlookdatabase.

noitacifinUnamreGfosnoitalumiSDOMITLUM .5elbaT ________________________________________________________________________________________________________________________ seirtnuoCMRErehtO ynamreGtseW ynamreGdetinU 3noitalfnI 1PDG laeR laeR 3noitalfnI 1PDG lareneG tnerruC 1PDG mreT-gnoL evitceffE tnemnrevoG tnuoccA tseretnI egnahcxE 2ecnalaB 2ecnalaB 3etaR 1etaR ________________________________________________________________________________________________________________________ oiranecSecnerefeR .1 0.0- 2.0- 7.0 5.1 2.0 6.0 2.2- 3.1- 6.0 0991 0.0- 1.0- 8.0 8.1 7.0 9.1 4.4- 7.2- 4.2 1991 1.0- 4.0- 8.0 0.2 2.0 7.0 7.2- 3.2- 2.4 49-2991 0.0 2.0- 3.0 1.1 2.0- 6.0 1.2- 9.1- 3.8 79-5991 3.0 2.0 1.0- 5.0- 2.0- 8.0 6.0- 9.0- 5.41 1002 oiranecScitsimitpOsseL .2 0.0- 2.0- 7.0 5.1 2.0 6.0 2.2- 3.1- 6.0 0991 0.0- 0.0- 8.0 8.1 8.0 0.2 8.4- 6.2- 6.1 1991 1.0- 3.0- 8.0 0.2 4.0 1.1 8.3- 6.2- 2.2 49-2991 0.0 2.0- 4.0 2.1 1.0- 4.1 7.3- 6.2- 4.4 79-5991 2.0 1.0 2.0 2.0- 1.0- 5.2 3.3- 3.2- 1.8 1002 ssoLytilibiderChtiwtnemngilaeRMRE .3 1.0- 2.0- 7.0 5.1 2.0 6.0 2.2- 3.1- 6.0 0991 3.1 6.0 6.0 6.2 5.0 9.1 4.4- 6.2- 4.2 1991 0.1 1.0- 7.0 8.1 3.0 7.0 6.2- 3.2- 2.4 49-2991 0.0 4.0- 4.0 7.0 2.0- 5.0 2.2- 8.1- 2.8 79-5991 4.0 2.0 2.0- 9.0- 2.0- 8.0 7.0- 6.0- 5.41 1002 ________________________________________________________________________________________________________________________ .5dna4selbaT,)b0991(htidereMdnanossaM :ecruoS .enilesabmorfnoitaivedtnecreP.1 .)dnahuerT(dnuFtsurTehtdnadnuFytinUehtfoticifedehtsedulcniecnalabtnemnrevoG .PDGenilesabfotnecrepninoitaiveD.2 .enilesabmorfnoitaivedtniopegatnecreP.3

noitacifinUnamreGfostcepsAfosnoitalumiSledoM2GSM .6elbaT ________________________________________________________________________________________________________________________ seirtnuoCMRErehtO ynamreGtseW ynamreGdetinU 3noitalfnI 1PDG laeR laeR 3noitalfnI 1PDG lareneG tnerruC 1PDG mreT-gnoL evitceffE tnemnrevoG tnuoccA tseretnI egnahcxE 2ecnalaB 2ecnalaB 3etaR 1etaR ________________________________________________________________________________________________________________________ )tnemngilaerMREon(noisnapxElacsiFnamreG .1 8.1- 0.2- 9.0 2.2 2.1- 8.0- 7.1- 2.1- .a.n 0991 5.1- 4.1- 3.1 0.4 3.0- 1.1 3.3- 2.2- 1991 1.0- 6.0- 2.1 8.3 0.0 0.1 3.3- 3.2- 49-2991 1.0 6.0- 2.1 4.3 1.0 9.0 4.3- 4.2- 79-5991 tnemngilaeRMREhtiwnoisnapxElacsiFnamreG .2 4.0 0.0 7.0 2.6 5.1- 0.1- 7.1- 8.1- .a.n 0991 5.0- 5.0- 1.1 0.7 3.0- 2.1 3.3- 8.2- 1991 1.0 2.0- 1.1 5.6 0.0 0.1 3.3- 9.2- 49-2991 2.0 3.0- 0.1 0.6 1.0 9.0 4.3- 1.3- 79-5991 )htworgfoetarniesaercni%2(noisnapxEyratenoMnamreG .3 8.0 9.0 1.0- 6.0- 7.0 9.0 3.0 2.0 .a.n 0991 4.1 4.1 3.0- 9.0- 3.1 4.1 4.0 1.0 1991 8.1 7.1 4.0- 1.1- 8.1 8.1 6.0 1.0 49-2991 9.1 0.2 5.0- 2.1- 9.1 1.2 6.0 1.0 79-5991 )yratenomdnalacsifgnirongi(noitargetnIlluFfostceffEediSylppuS .4 8.1 5.2 2.1 0.11- 1.2- .a.n 0.1- 7.2- 8.0 0991 9.0 5.1 9.0 2.21- 6.1- 3.0- 8.2- 4.3 1991 3.0 1.1 7.0 3.41- 6.0- 3.0 0.3- 5.5 49-2991 2.0 0.1 6.0 4.61- 3.0- 8.0 1.3- 2.7 79-5991 ________________________________________________________________________________________________________________________ .4hguorht1selbaT,)0991(nibbiKcM :ecruoS .enilesabmorfnoitaivedtnecreP.1 .PDGenilesabfotnecrepninoitaiveD.2 .enilesabmorfnoitaivedtniopegatnecreP.3

noitacifinUnamreGfosnoitalumiS3XM .7elbaT ________________________________________________________________________________________________________________________ seirtnuoCMRErehtO ynamreGtseW ynamreGdetinU 3noitalfnI 1PDG laeR laeR 3noitalfnI 1PDG lareneG tnerruC 1PDG mreT-gnoL evitceffE tnemnrevoG tnuoccA tseretnI egnahcxE 2ecnalaB 2ecnalaB 3etaR 1etaR ________________________________________________________________________________________________________________________ )MREon(oiranecSecnerefeR .1 0.0 0.0 4.0 1.5 1.0 .a.n 0.3- 3.0 2.0 0991 0.0 0.0 8.0 7.9 2.0 9.4- 9.0- 1.1 1991 0.0 0.0 7.0 3.8 0.0 0.1- 6.2- 4.5 49-2991 0.0 1.0- 6.0 4.6 0.0 4.1 5.3- 8.21 79-5991 )MDotdeggepWOR(oiranecSMRE .2 7.1- 3.1- 6.0 0.2 1.0 .a.n 1.3- 0.0 2.0 0991 7.5- 9.3- 2.1 8.6 1.0 1.5- 6.1- 9.0 1991 1.1- 0.3- 4.1 2.11 2.0 0.1- 6.1- 1.5 49-2991 1.2 1.1- 2.1 0.9 2.0 6.1 0.1 3.21 79-5991 yciloPyratenoMnamreGevitadomoccAhtiwoiranecSMRE .3 2.1- 9.0- 5.0 2.2 4.0 .a.n 1.3- 1.0 4.0 0991 9.3- 0.3- 8.0 0.7 3.1 3.5- 3.1- 6.1 1991 8.0- 8.1- 9.0 1.11 5.0 2.1- 8.1- 8.5 49-2991 8.0 6.0- 7.0 6.8 4.0- 4.1 5.0- 0.31 79-5991 ________________________________________________________________________________________________________________________ .9Adna,7A,1AselbaT,)3991(nongaGdna,rednaxelA,smadA :ecruoS .enilesabmorfnoitaivedtnecreP.1 .)dnahuerT(dnuFtsurTehtdnadnuFytinUehtfoticifedehtsedulcniecnalabtnemnrevoG .PDGenilesabfotnecrepninoitaiveD.2 .enilesabmorfnoitaivedtniopegatnecreP.3

49-9891,selbairaVcimonoceorcaMdetceleS :ynamreG .8elbaT )tnecrepnI( ________________________________________________________________________________________________________________________ seirtnuoCUErehtO ynamreGtseW ynamreGdetinU noitalfnI PDG -iMteN laeR laeR noitalfnI PDG lareneG tnerruC tuptuO PDG htworG noitarg mreT-gnoL evitceffE htworG tnemnrevoG tnuoccA 1paG htworG tsaEmorf tseretnI egnahcxE 2ecnalaB 2ecnalaB 4ynamreG etaR 3etaR ________________________________________________________________________________________________________________________ 3.5 5.3 3.383 2.4 -- 8.2 6.3 1.0 8.4 -- 6.3 9891 0.6 2.2 1.953 0.6 3.5 7.2 7.5 9.1- 0.3 0.2 7.5 0991 5.5 7.0 5.961 0.5 2.4 5.3 0.5 3.3- 1.1- 2.3 8.2 1991 4.4 8.0 8.78 9.3 6.7 0.4 8.1 9.2- 1.1- 8.1 2.2 2991 5.3 3.0- 3.35 5.2 5.51 1.4 7.1- 3.3- 8.0- 2.2- 2.1- 3991 1.3 8.2 3.72 9.3 9.71 0.3 3.2 5.2- 0.1- 2.2- 9.2 4991 ________________________________________________________________________________________________________________________ .sesabatadsrotacidnIcimonocEtnerruCdnakooltuOcimonocEdlroW :secruoS .stniopegatnecrepni,eulav9891morfnoitaivedtuptuolaitnetop/lautcA.1 .PDGotoitaR.2 .eulav9891morfnoitaivedtnecreP.3 .59/1dna59/21kitsitatSdnutfahcstriW,tmasednuBsehcsitsitatS :ecruoS .snosrepfosdnasuohtnI.4