1、 通貨膨脹動態(tài)和通脹預期外文翻譯2019-2020英文GlobalinflationdynamicsandinflationexpectationsMartinFeldkirchera,PierreSiklosAbstractInthispaperweinvestigatedynamicsofinflationandshort-runinflationexpectations.Weestimateaglobalvectorautoregressive(GVAR)modelusingBayesiantechniques.Wethenexploretheeffectsofthreesourceof

2、inflationarypressurethatcoulddriveupinflationexpectations:domesticaggregatedemandandsupplyshocksaswellasaglobalincreaseinoilpriceinflation.Ourresultsindicatethatinflationexpectationstendtoincreaseasinflationaccelerates.However,theeffectsofthedemandandsupplyshocksareshort-livedformostcountries.Whengl

3、obaloilpriceinflationaccelerates,however,effectsoninflationandexpectationsareoftenmorepronouncedandlong-lasting.Hence,anassessmentofthelinkbetweenobservedinflationandinflationexpectationsrequiresdisentanglingtheunderlyingsourcesofinflationarypressure.Wealsoexaminewhethertherelationshipbetweenactuali

4、nflationandinflationexpectationschangedfollowingtheglobalfinancialcrisis.Thetransmissionbetweeninflationandinflationexpectationsisfoundtobelargelyunaffectedinresponsetodomesticdemandandsupplyshocks,whileeffectsofanoilpriceshockoninflationexpectationsaresmallerpost-crisis.Keywords:Inflation,Inflation

5、expectations,GVARmodelling,AnchoringofinflationexpectationsIntroductionInflationexpectationsareapivotalvariableinprovidinginsightsaboutlikelyfutureeconomicconditions.Whilethedecadeslongdebateaboutthedegreetowhichmonetarypolicyisforwardlookinghasnotabated(e.g.,Friedman,1968;Woodford,2003a)thereislitt

6、ledoubtthatpolicymakersdevoteconsiderableattentiontotheeconomicoutlook.Hence,thedynamicsoftherelationshipbetweeninflationandinflationexpectationscontinuestopre-occupythemonetaryauthoritiesandcentralbankers.Evenbeforethefullimpactoftheglobalfinancialcrisis(GFC)of20089wasfeltintheUS,andinmanyotherpart

7、softheglobe,centralbankerssuchasBernanke(2007)highlightedtheimportanceofinflationexpectationssince“thestateofinflationexpectationsgreatlyinfluencesactualinflation“.Morerecently,Yellen(2016)alsounderscoresthecrucialroleplayedbyexpectationswhilebemoaningthefactthattheprofessionmustconfrontgapsinourkno

8、wledgeabouttherelationshipbetweenobservedinflationandtheshort-runinflationexpectationsthatliesattheheartofmanytheoreticalmacroeconomicmodels.Itisnotdifficulttocomeacrossspeechesbycentralbankerswho,onaregularbasis,touchuponthesubjectoftheformationandimplicationsofinflationexpectations.Amain,butnotsol

9、e,driverofinflationexpectationsispastinflation.Attheriskofsomeover-simplification,inflationcanbethoughtofasbeingdrivenbytwosetsofdeterminants,namelylocalordomesticfactorsversusinternationalorglobalforces.Thelocaldeterminantswouldincludetechnicalprogressandchangesinproductivity,demographicfactors,ins

10、titutionalconsiderationssuchastheadoptionofinflationtargetingandcentralbankindependenceand,since2008,theadoptionandmaintenanceofunconventionalmonetarypoliciesinsystemicallyimportanteconomies.Moregenerally,however,economiststendtomakethedistinctionbetweenaggregatedemandandsupplysourcesofchangesininfl

11、ationpressure.Inwhatfollows,weretainthisdistinctiontoallowforgreatercomparabilitywiththeextantliteratureaswellasbecauseitprovidesuswithavehicletopresentnewinsightsintotheunderlyingdriversofinflationandultimatelyaboutthelikelihoodthatinflationexpectationscanbeanchored.Globalizationinboththetradingofg

12、oodandservicesandinfinanceisoftenalsotoutedasacriticaldriveroftheinternationalcomponentthatinfluencesdomesticinflationrates.Asaresult,theextantliteraturehasdivergedwhereinsomearguethatmodelsofinflationaretoonationcentric(e.g.,Borio&Filardo,2007;Ahmad&Civelli,2016;Auer,Borio,&Filardo,2017,Kabukguolu&

13、Martinez-Garcia,2018)whileothersplacegreateremphasisonthevariouslocalfactorsmentionedabove.Thecurrentliteraturegenerallyfocusesonahomogeneoussetofcountries(e.g.,advancedoremergingmarketeconomies;seethefollowingsection).Wedepartfromthisnormtoconsider42economiesthatspanawiderangeintermsoftheirsize,suc

14、cessatcontrollinginflation,monetarystrategiesinplace,andtheextenttowhichtheyweredirectlyimplicatedornotintheGFC.Tofullyexploitthepotentialforcross-borderspilloversininflationweusetheGlobalVARs(GVAR)methodology(Pesaran&Chudik,2016).Thismethodologyiswellsuitedtoaddressthedomesticimpactofchanginginflat

15、iononexpectationsdynamicscontrollingforinternationalspilloversthroughcross-borderinter-linkages.Often,globalfactorsareconstructedusingsomeindicatoroftradeopennesstoaggregatecountry-specificseries.Nevertheless,thereisdisagreementaboutwhetherthisistheappropriatevehicletoestimateglobalversuslocalinflue

16、ncesondomesticinflation(e.g.,Kabukguo.iu&Martinez-Garcia,2018;Ahmad&Civelli,2016).Instead,weproposeanovelsetofweightsinestimatingtheGVARobtainedfromtheforecasterrorvariancedecompositionsestimatedviathemethodologyofDieboldandYilmaz(DY,2009)developedtomeasurethedegreeofconnectedness.Sincethedebateabou

17、tlocalversusglobaldeterminantsofinflationexpectationspartlycentersaroundtheextenttowhichcountriesarelinkedtoeachothertheDYtechniqueisanaturalonetouseinthepresentcircumstances.Indeed,theforegoingcombinationofmethodologiespermitsustohighlighttwoneglectedaspectsofthedebateaboutwhatdrivesinflationandinf

18、lationexpectations.First,thattherelativeimportanceoflocalversusglobalfactorsislikelyafunctionofthepolicyhorizon.Second,theglobalizationofobservedinflationisalsoreflectedinaglobalizationofexpectedinflation.WhiletheGVARmethodologyprovidesaveryrichsetofpotentialshocksthatmaybeanalyzed,wefocusontwosetso

19、fshocks.Theyare:theimpactofdomesticaggregatedemandandsupplyshocksoninflationandinflationexpectationsandtheimpactofaglobaloilpricesupplyshockonthesesametwovariables.Briefly,wefindthatinflationexpectationsrespondpositivelytoeitherdomesticaggregatedemandorsupplyshocks,buttheeffectsaregenerallytemporary

20、.Thisfindingholdsequallytrueforthepost-crisisperiod.Bycontrast,ifinflationacceleratesduetoapick-upinglobaloilpriceinflation,inflationexpectationsrespondsignificantlypositiveandeffectsarelonglasting.Theimpactoninflationisevenlargerthanoninflationexpectations.Hence,oilpriceshocksdriveawedgebetweeninfl

21、ationandinflationexpectationsevenamongprofessionalforecasters.Nevertheless,actualinflationandinflationexpectationstendtoco-movecloselyandthepass-throughhasdiminishedintheaftermathofthecrisis.Therefore,inanerawhereenergypricesarevolatileandaresubjecttolargeswings,thishasimplicationsforwhenandhowaggre

22、ssivelymonetaryauthoritiesoughttorespondtooilpriceshocks.Anadditionalimportantimplicationisthatidentifyingtheaggregatesupplyfromaggregatedemandcomponentsofshocksiscriticaltounderstandingthedynamicsofbothobservedandexpectedinflation.LiteraturereviewInflationexpectationslieatthecoreofallmacroeconomicm

23、odels(e.g.,Woodford,2003a).Moreover,totheextentthatpolicyisabletoinfluencetheseexpectations,understandingtheconnectionwithobservedinflationremainsanessentialingredienttoevaluatingtheimpactofmonetarypolicy.EspeciallyfollowingtheGFC,thedebatesurroundingthemechanismthatbestdescribeshowexpectationsadjus

24、tinresponsetoshocks,aswellaswhatarethefundamentaldriversofinflationexpectations,hasbeenrekindled.Thesameistrueofthecompanionliteraturethatexploresthedynamicsanddeterminantsofobservedinflation.Aneraofultra-lowinterestrates,combinedwithlowinflation,hasalsocontributedtorevivingthestudyanddebateaboutlin

25、ksbetweeninflationandinflationexpectations.Rationalexpectationsserveasaconvenientbenchmark,inpartbecausetheoreticalmodelsarereadilysolvableandclosedformsolutionsaretypicallyfeasible.However,whenconfrontedwiththeempiricalevidence,considerabledifferencesofopinionemergeabouthowbesttodescribetheevolutio

26、nofexpectations.Forexample,anearlyassessmentbytheBankofJapanofitsQuantitativeandQualitativeEasingprogram(QQE;BankofJapan,2016)findsthattheJapanesearepronetoadjustinginflationexpectationsmoregraduallythaninotheradvancedeconomies(e.g.,theUSortheeuroarea).Thisislargelyduetothebackward-lookingnatureofth

27、eseexpectations.ThisalsoresonatessomewhatwithrecentevidencefromtheUS(e.g.,Trehan,2015)andothereconomiesbothlargeandsystemicallyimportantaswellasonesthataresmallandopen(e.g.,Bhatnagaretal.,2017).Ofcourse,theremaybeseveralexplanationsforthesluggishadjustmentofinflationexpectations.Japan,aftertwodecade

28、sormoreofverylowinflationtolowdeflation,setsthiscountryapartfromtheremainingadvancedeconomieswhich,overthesameperiod,experiencedonlypassingboutsofdeflation(early2000sandintheaftermathofthe2007-8globalfinancialcrisis).Sincethattime,belownormalinflationrateshavespreadacrossmuchoftheadvancedworld.Unsur

29、prisingly,thishasattractedtheattentionandtheconcernofpolicymakers.Thisrepresentsarelativelynewelementinthestoryofthedynamicsofinflation.6Itisalsonotablethat,priortotherecentdropininflation,themainconcernwastheroleofcommodityprices,notablyoilprices,ingeneratinghigherinflationandtheextenttowhichtheses

30、hockswereseentohavepermanenteffectsornot.Evenifdomesticeconomicslackretainsitspowertoinfluenceinflation,theglobalizationoftradeandfinancehasintroducedanewelementintotheinflationstory,namelythepotentialroleofglobalslack.Rogoff(2003,2006)earlyondrewattentiontothelinkbetweenthephenomenonofglobalization

31、andinflation.Alternatively,atalmostthehighpointoftheglobalizationera,studiesbegantoappearthatprovidedempiricalsupporteitherinfavorofasignificantglobalcomponentininflation,insomeofitscriticalcomponents(e.g.,Ciccarelli&Mojon,2010;Parker,2017),orviatheglobalinfluenceofChinasrapideconomicgrowth(e.g.,Pan

32、g&Siklos,2016,andreferencestherein).Economists,centralbankersandpolicymakershavewaxedandwanedintheirviewsaboutthesignificanceofglobalslackasasourceofinflationarypressure.Nevertheless,itisaconsiderationthatneedstobetakenseriouslyandthequestionremainsunderstudied(e.g.,seeBorio&Filardo,2007,Ihrig,Kamin

33、,Lindner,&Marquez,2010andYellen,2016).Morebroadly,thenotionthataglobalcomponentisanimportantdriverofdomesticinflationratescontinuestofindempiricalsupportdespiteoftheproliferationofneweconometrictechniquesusedtoaddressthequestion(e.g.,Carriero,Corsello,&Marcellino,2018).Recallingthewordsofcentralbank

34、erscitedintheintroductionthereremainsmuchtobelearnedaboutthedynamicrelationshipbetweenobservedandexpectedinflation.Thetwoareinextricablylinked,forexample,intheorybecausetheanchoringofexpectationsisthoughttobethecorerequirementofasuccessfulmonetarypolicystrategythatpreventsprices(andwages)fromdriftin

35、gawayeitherfromastatedobjective,asininflationtargetingeconomies,oranimplicitonewherethecentralbankiscommittedtosomeformofpricestability.Thereis,ofcourse,alsoanever-expandingliteraturethatexamineshowwellexpectationsareanchored.Thisliteraturefocusesmostlyonlong-runinflationexpectations(e.g.,seeBuono&F

36、ormai,2018;Chan,Clark,&Koop,2017;Mehrotra&Yetman,2018;Lyziak&Paloviita,2016;Strohsal&Winkelmann,2015,andreferencestherein).Afewauthorshavefocusedonepisodeswheninflationisbelowtarget(e.g.,Ehrmann,2015),orduringmilddeflations(Banerjee&Mehrotra,2018),whileIMF(2016),Blanchard(2016),andBlanchard,Cerutti,

37、andSummers(2015),aremoregeneralinvestigationsoftheissues.Whilecentralbankersandaconsiderableportionofexistingempiricalresearchworriesabouthowchangestoinflationexpectationsinfluenceobservedinflationthemorerecentliteratureontheanchoringofexpectationsshiftstheemphasisonhowinflationshockscande-anchorthe

38、sesameexpectations.Unfortunately,thereisasyetnoformaldefinitionofanchoring.Indeed,thereisstillnothingapproachingaconsensusonthedeterminantsofinflationexpectationsfromvarioussources(e.g.,households,firms,professionals).Factorsrangefromthepasthistoryofinflation,knowledgeofmonetarypolicy,mediaportrayal

39、softheinflationprocess,shoppingexperience,andtheimpactofcommodityprices,tonamesomeofthemostprominentdeterminants(e.g.,seeCoibion,Gorodnichenko,Kumar,&Pedemonte,2018b).Notlistedisaroleforglobalfactorswhichisafocusofthepresentstudy.Since,asweshallseebelow,thereisacloseconnectionbetweeninflationandexpe

40、ctedinflation,andanincreasinglywell-establishedlinkbetweenglobalandlocalinflation,thereexistsanadditionalunexploredavenuethattiesinflationperformanceonaglobalscaletolocalinflationexpectations.Wecontributetothisliteratureinthesensethatwealsoquantifytheeffectsofinflationaryshocksonshort-runinflationex

41、pectations,whichhasabearingontheabilityofcentralbankstoanchorinflationinthemedium-term.Adhocmethodsexisttoconvertfixedeventintofixedhorizonforecasts(e.g.,seeBuono&Formai,2018;Siklos,2013).Winkelried(2017)adaptstheKozickiandTinsley(2012)shiftingendpointmodeltoexploittheinformationcontentoffixedeventf

42、orecasts.Ofcourse,wedonotknowwhetherorhowmuchnewinformationisabsorbedintosubsequentforecastsinanenvironmentwherethehorizonshrinkswhetheritisbecauseinformationisstickyorthereissufficientrationalinattentionthatmitigatestheeffectivedifferencesbetweenfixedhorizonandfixedeventforecasts.Althoughconstructe

43、dfixedhorizonforecastsareimperfect(e.g.,Yetman,2018)theyhavetheadvantagethatseveralpapersintheextantliteratureemploythisproxy.Untiltherecentperiodofsluggishinflation,thefocusofmuchresearchfellonaccountsthatsoughttoevaluatethesuccess,orlackthereof,ofinflationtargeting(IT)regimes.Thefactthatthiskindof

44、monetarypolicystrategywasdesignedinanerawherethechallengewastoreduceinflationisnotlostonthosewhoaskwhetherITregimesareuptothetaskofmaintaininginflationclosetothetarget(e.g.,seeEhrmann,2015andreferencestherein).Fuhrer(2017)considerstheextenttowhichexpectationsofinflationareinformativeaboutthedynamics

45、ofobservedinflationbasedonempiricalworkthatcoversaperiodof25yearsfortheUSandJapan.Fuhrersstudyisalsointerestedintheextenttowhichlong-terminflationobjectivescanbemodelledviaasequenceofshort-termforecasts.Theanswerseemstobeintheaffirmativebutsignificantdeparturesfromthelong-termarepresentinthedata.Inc

46、ontrast,ourstudyisnotabletodeterminethestrengthofanysuchlinksduetodatalimitations,asweshallsee.Nevertheless,assuggestedabove,thedegreetowhichinflationisanchoredneednotbesolelyevaluatedaccordingtolong-termexpectations.Short-termdeviationscanalsoserveaswarningsignals.Toourknowledgethen,adynamicmodelth

47、atattemptstoevaluatethelinkbetweeninflationandinflationexpectationsineconomiesbeyondonesthatareadvanced,andtheroleofglobalfactorsaswellascross-countryinteractions,isstillmissing.Thefollowingsectionsbegintofillthegap.EmpiricalresultsWebeginbyinvestigatingtheresponsesofinflationexpectationsandactualin

48、flationtoadomesticADshock.Weseethatformosteconomiestheresponseofinflationexpectationsiseitherflatandhoversaroundzeroorishumpshaped,peteringoutinthelongerterm.Thisfindingimpliesahighdegreeofanchoringofshort-terminflationexpectations,whichmightdirectlytranslateintoanchoringoflong-runinflationexpectati

49、ons.Countriesforwhichinflationexpectationsconvergemoreslowlycompriseadvancedandeuroareaeconomies(Italy,Ireland,Norway,PortugalandSlovenia),CESEEeconomies(Bulgaria,Croatia,Russia),Asianeconomies(China,IndiaandIndonesia)aswellasSouthAfrica.InItaly,Norway,SouthAfricaandSloveniathecoolingoffphaseofinfla

50、tionexpectationstakesparticularlylong.InflationexpectationsdecreaseinIndia,IndonesiaandChile.Doinflationexpectationsandactualinflationalwaysmoveinthesamedirection?Weseethatactualinflationresponsesarenothumpshapedratherinmostcountriesactualinflationgraduallydeclinesanddiesoutafter816months.However,in

51、countriesthatshowalongeradjustmentphase,actualinflationresponsesalsotakelongertocooloff.IncountrieswithnegativeinflationexpectationresponsesactualinflationtendstobenegativeovertheimpulseresponsehorizonsuchasinthecasesofIndiaandChile.InIndonesia,bycontrast,actualinflationispositiveupuntil20monthsreve

52、alinganegativerelationshipbetweeninflationexpectationsandinflation,which,however,isnotstatisticallysignificantsincecredibleintervalsofbothresponsesoverlap.Asignificantlylargerlong-runeffectofinflationcomparedtoinflationexpectationscanbefoundinPeru,though.Next,weinvestigatetheimpactofdomesticASshocks

53、oninflationexpectations.Here,wealsofindthatinflationexpectationstendtoadjustratherquicklyforalmostallcountries.Shapesofactualinflationresponsestendtodifferfromthoseforinflationexpectationsindicatingthatthereisnodirectone-to-onerelationshipbetweenthetwoseries.Exceptionstothisresultareobtainedincountr

54、iesthateithershowapositiveandsignificantresponseofinflationexpectationsinthelong-run(BulgariaandCroatia)oranegativeresponse(India,BrazilandChile).Inthesecountries,inflationexpectationstendtofollowcloselyactualinflationresponsesimplyingthatthereisnodivergencebetweenresponsesofinflationexpectationsand

55、actualinflation.Onlyintwocountriesaredifferencesinlong-runresponsesofinflationexpectationsandinflationstatisticallysignificant.Morespecifically,inChile,responsesofinflationexpectationsaresignificantlymorepronouncedthanresponsesofactualinflation,whereastheoppositeisthecaseinPeru.Finally,welookattheef

56、fectsofasupplysidedrivenaccelerationofoilpriceinflation.Here,weseethatmostoftheeffectsarepositiveandlong-lasting.Also,theeffectsaresometimesrathersizableeveninthelong-run.Thebottompanelshowsdifferencesininflationexpectationsandinflationresponsesalongwith68%credibleintervals.Thefigureindicatesthateve

57、nafter25months,effectsonactualinflationaresizableandforsomecountriessignificantlylargerthanthoseoninflationexpectations.Regressingtheposteriormedianofthesedifferencesonthesumofexportsandimportsin%ofGDP,averagedoverthesampleperiod,showsthatcountrieswithahigherdegreeoftradeopennessshowlargerdifference

58、sinresponses.Theimplicationisthatformoreopeneconomiestheeffectofoilpricechangesondomesticinflationislarger,therebydrivingawedgebetweenexpectationsandactualinflation.ThattheeffectofglobalshocksondomesticinflationislargerformoreopeneconomiesisconsistentwiththefindingsprovidedinAhmadandCivelli(2016).Ha

59、vingsaidthat,responsesofinflationexpectationsinabsolutetermsalsotendtobesizeable.Indeed,in23of42countries,longrun-effectsoninflationexpectationsareabove0.4.Thisimpliesthatnearlyhalfoftheaccelerationinoilpricesdirectlytranslatesintoupwardmovementsofinflationexpectations.Incontrasttothedomesticdemanda

60、ndsupplyshocks,therearenosignificantlynegativeresponsesforthecountriescoveredinthisstudy.Finally,comparingtheshapesofimpulseresponses,wefindastrongrelationshipbetweenactualinflationandinflationexpectationresponses.Summingup,inflationexpectationsincreaseintheshort-runifinflationarypressurestemsfromei