商业周期的形状：弗里德曼拔毛理论的跨国分析（英）

BISWorkingPapers No1076 Theshapeofbusinesscycles:across-countryanalysisofFriedman’spluckingtheory byEmanuelKohlscheen,RichhildMoessnerandDanielMRees MonetaryandEconomicDepartment February2023 JELclassification:E24;E32. Keywords:businesscycle,growth,labourmarket, unemployment. BISWorkingPapersarewrittenbymembersoftheMonetaryandEconomicDepartmentoftheBankforInternationalSettlements,andfromtimetotimebyothereconomists,andarepublishedbytheBank.Thepapersareonsubjectsoftopicalinterestandaretechnicalincharacter.TheviewsexpressedinthemarethoseoftheirauthorsandnotnecessarilytheviewsoftheBIS. ThispublicationisavailableontheBISwebsite(www.bis.org). ©BankforInternationalSettlements2023.Allrightsreserved.Briefexcerptsmaybereproducedortranslatedprovidedthesourceisstated. ISSN1020-0959(print) ISSN1682-7678(online) Theshapeofbusinesscycles:across-countryanalysisofFriedman’spluckingtheory EmanuelKohlscheen,RichhildMoessnerandDanielM.Rees1 Abstract WetesttheinternationalapplicabilityofFriedman’sfamouspluckingtheoryofthebusinesscyclein12advancedeconomiesbetween1970and2021.Wefindthatincountrieswherelabourmarketsareflexible(Australia,Canada,UnitedKingdomandUnitedStates),unemploymentratestypicallyreturntopre-recessionlevels,inlinewithFriedman’stheory.Elsewhere,unemploymentratesarelesscyclical.Outputrecoveriesdifferlessacrosscountries,butmoreacrossepisodes:onaverage,halfofthedeclineinGDPduringarecessionpersists.Intermsofsectors,declinesinmanufacturingaretypicallyfullyreversed.Incontrast,construction-drivenrecessions,whichareoftenassociatedwithburstingpropertypricebubbles,tendtobepersistent. JELClassification:E24;E32. Keywords:businesscycle;growth;labourmarket;unemployment 1BankforInternationalSettlements,Centralbahnplatz2,4002Basel,Switzerland.WearegratefultoEgonZakrajšekforhelpfulcommentsandsuggestionsandEmeseKurucforexcellentresearchassistance.TheviewsexpressedinthispaperarethoseoftheauthorsanddonotnecessarilyreflectthoseoftheBankforInternationalSettlements. 1.Introduction Areeconomicrecoveriesfromrecessionscomplete?Ifnot,whatcharacteristicsinfluencetheextentofrecoveries?Inthisarticle,wepresentnovelcross-countryevidenceontheshapeofbusinesscyclesacrossadvancedeconomies.WecomparethepatternsthatarefoundwiththepredictionsofMiltonFriedman’s“pluckingtheory”ofthebusinesscycle(seeFriedman(1964,1993)).ThistheorywasrecentlyrevivedbytheinfluentialpaperofDuprazetal(2019).2 Ouranalysisisbasedontwelveadvancedeconomies(AEs),coveringatotalofmorethan50recessionssince1970.3Recessionsineachcountryaretimedusingthesametimingalgorithm,followingBryandBoschan(1971)andHardingandPagan(2003).Ourcentralfindingisthatincountrieswithflexiblelabourmarketspost-recessiondeclinesintheunemploymentratealmostfullyreversetheeffectsofrecessions.Inotherwords,recoveriesinthesefollowafairlypredictablepattern.Further,theunemploymentratedeclinesinexpansionsbearnorelationwithrisesinsubsequentrecessions.ThispatternisbroadlyconsistentwithFriedman’spluckingtheoryofthebusinesscycle. Yet,inthetwothirdsofcountrieswithlessflexiblelabourmarketsunemploymentratesarefarlesscyclical.Incontrast,outputrecoveriesdifferlessacrosscountriesbutmoreacrossepisodes.Manyrecessionsleavepermanentscars.Itisimportanttonoteherethat,throughout,wefocusonthemagnitudeofoutputandemploymentdeclinesandsubsequentrecoveries,ratherthanlookingatthetimingorcauseofturnsinthebusinesscycle.4 2.Datingbusinesscyclesacrosscountries Figure1showsthetimelineofalleconomicexpansionsintwelveAEssince1970.ExpansionsandrecessionsweretimedbyapplyingtheBry-Boschan-Harding-PaganproceduretoquarterlytimeseriesofseasonallyadjustedrealGDP.5Forcountrieswherecommonly-acceptedrecessiondatesexist(asintheUnitedStates),thetimelinealignscloselywiththosesources. 2KimandNelson(1999)statethatinFriedman’spluckingview,“recessionsarelikecommoncold:theycomeonsuddenlyandrecoveryfollowsafairlypredictablecourse,butthetimethathaspassedsincethelastcoldisofnouseinpredictingwhenthenextwilloccur,oritsseverity.” 3Australia,Canada,France,Germany,Italy,Japan,Norway,Spain,Sweden,Switzerland,theUnitedKingdomandtheUnitedStates.Countryselectionwasbasedsolelyondataavailability(overlongspan). 4Whereastheoverallstateofaneconomymaynotprovideahintofwhetheranewrecessionislikely,otherindicators,suchastheslopeoftheyieldcurveordownturnsinresidentialconstructionactivityhavebeenshowntoanticipatethetimingofturningpoints.Seee.g.EstrellaandHardouvelis(1991)andKohlscheenetal(2020). 5AlldataweresourcedfromtheOECD.Foracompleteexplanationofthetimingalgorithm,seeHardingandPagan(2003).Aminimumoftwoquartersperphasewasimposed. Figure1Timelineofeconomicexpansions1 PeaksandthroughsofbusinesscyclesastimedbyBry-Boschan-Harding-Paganalgorithm 1ThegraphisshowingexpansionperiodssinceQ11970,exceptforCA(Q31970),CH(Q31981),DE(Q31991),ES(Q41980),FR(Q31974),IT(Q41981),JP(Q41973)andNO(Q11980). Onaverage,wefindthatbusinesscyclesinadvancedeconomieslast26quarters(6½years)althoughthisvariessignificantlyacrosscountriesandovertime.Theidentifieddatesalsopointtosignificantbusinesscycleasymmetry:e