IMF-高負債是否制約貨幣政策？來自通脹預(yù)期的證據(jù)（英）

IsHighDebtConstrainingMonetaryPolicy?EvidencefromInflationExpectationsWP/23/143IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.2023JUN?2023InternationalMonetaryFundWP/23/143IMFWorkingPaperMonetaryandCapitalMarketsandWesternHemisphere

DepartmentsIsHighDebtConstrainingMonetaryPolicy?Evidencefrom

InflationExpectationsPreparedbyLuisBrandao-Marques,MarcoCasiraghi,GastonGelos,OlamideHarrison,andGunesKamber*AuthorizedfordistributionbyDavidHofmanJune2023IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andare

publishedtoelicitcommentsandtoencouragedebate.

TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.ABSTRACT:Thispaperexamineswhetherhighgovernmentdebtlevelsposeachallengetocontaininginflation.Itdoessobyassessingtheimpactofgovernmentdebtsurprisesoninflationexpectations

inadvanced-andemergingmarketeconomies.

Itfindsthatdebtsurprisesraise

long-terminflationexpectationsinemergingmarketeconomiesin

apersistentway,butnotinadvancedeconomies.Theeffectsarestrongerwheninitialdebtlevelsarealreadyhigh,wheninflationlevelsareinitiallyhigh,andwhendebtdollarizationissignificant.

Bycontrast,debtsurpriseshaveonlymodesteffectsin

economieswithinflationtargetingregimes.Increaseddebtlevelsmaycomplicatethefightagainstinflationin

emergingmarketeconomieswithhighanddollarizeddebtlevels,andweakermonetarypolicyframeworks.RECOMMENDEDCITATION:JELClassificationNumbers:Keywords:E31,E41,E52,E62Inflationexpectations;monetarypolicy;fiscaldominance;debtAuthor’sE-MailAddress:lmarques@,mcasiraghi@,ggelos@,oharrison@;gkamber@IMFWORKINGPAPERSDebtSurprisesandInflationExpectationsWORKINGPAPERSIsHighDebtConstrainingMonetary

Policy?EvidencefromInflation

ExpectationsPreparedby

LuisBrandao-Marques,MarcoCasiraghi,GastonGelos,OlamideHarrison,andGunesKamber11

Theauthorswouldliketothank

GustavoAdler,BergljotBarkbu,DavidHofman,andErlendNierfortheircomments.ZoeStraussprovidedexcellentresearchassistance.INTERNATIONALMONETARYFUND3IMFWORKINGPAPERSDebtSurprisesandInflationExpectationsContentsINTRODUCTION__________________________________________________________________________________

5THEORETICALMECHANISMS

__________________________________________________________________

6DATAANDEMPIRICALMETHODS

_____________________________________________________________

7EconometricApproach____________________________________________________________________________

7Data

______________________________________________________________________________________________

10RESULTS________________________________________________________________________________________

12MainResults–EmergingMarket-vs.AdvancedEconomies

____________________________________

12StateDependence

_______________________________________________________________________________

15Government

DebtLevel

_________________________________________________________

15InflationLevel

__________________________________________________________________

17DebtDollarization_______________________________________________________________

19Inflation-Targeting________________________________________________________________________________

20ROBUSTNESS

__________________________________________________________________________________

22CONCLUSION

___________________________________________________________________________________

24REFERENCES___________________________________________________________________________________

25FIGURES1.DataAvailability

_______________________________________________________________

112.Debt-to-GDPSurprisesonImpact

_________________________________________________

113.DebtSurprisesAcrossWEOVintages______________________________________________

124.Responseof5-YearInflationExpectationstoGovernmentDebt

Shocks,Baseline

___________

145.Responseof5-YearInflationExpectationstoGovernmentDebt

Shocks,byInitialDebt

_______

166.Responseof5-YearInflationExpectationstoGovernmentDebt

Shocks,byInitialInflationLevelClassification

_________________________________________________________________

187.Responseof5-YearInflationExpectationstoGovernmentDebt

Shocks,EmergingMarkets—InitialFXShareofGovernmentDebt_______________________________________________________

208.ResponseofInflationExpectationstoGovernmentDebtShocks,byInflationTargetingRegimeClassification

_________________________________________________________________

219.ResponseofInflationExpectationstoGovernmentDebtSurprises,EmergingMarkets—ConditionalontheFiscalStance

______________________________________________________________

2210.ResponseofInflationExpectationstoGovernmentDebtSurprises,Emerging______________

23TABLES1.SummaryStatistics

____________________________________________________________

102.Responseof5-YearInflationExpectations,Baseline(FullSample)

_______________________

143.Responseof5-YearInflationExpectations,Baseline(EmergingMarkets)

__________________

144.Responseof5-YearInflationExpectations,Baseline(AdvancedEconomies)

_______________

155.Responseof5-YearInflationExpectations,byInitialDebtLevelClassification

______________

167.Responseof5-YearInflationExpectations,byInitialInflationLevel_______________________

188.Responseof5-YearInflationExpectations,byInitialFXDebtShareClassification

___________

209.RegressionResults,byICRGFinancialRiskRating,EmergingMarkets

___________________

24ANNEXESANNEXI

__________________________________________________________________________________________

29ANNEXII

_________________________________________________________________________________________

30INTERNATIONALMONETARYFUND4IMFWORKINGPAPERSDebtSurprisesandInflationExpectationsIntroductionTheCOVID-19crisispropelledsovereigndebtlevelsaroundtheworldtonewheights.Between2019and2022,inadvancedeconomies,debtlevelsrosefrom103.9to112.5percentofGDP,andinemergingmarketeconomies,theygrewfrom55.1to64.6percentofGDP.Sovereigndebtlevelsareexpectedtoremainhighincomingyears(IMF,2022),anddemandsonfiscalpolicyremainsubstantial.Atthesametime,higherlevelsofinterestratescouldincreasedebtservicesconsiderablyinthecomingyears.Couldtheriseingovernmentdebtlevelsposeaproblemformonetarypolicy?Thisquestionisofparticularrelevanceascentralbanksarestrugglingwiththemajorchallengeposedbyinflationarypressures.Inprinciple,variousmechanismsby

whichhighsovereigndebtlevelscancomplicatemonetarypolicymakingareconceivable.Centralbanksmaybe(seenas)hesitanttoraiseinterestratesasmuchasneededtoachievetheirinflationobjective,outofconcernfordebtsustainability.Theymayalsobeconcernedabouttheimpactofrisingratesontheirownnetincome

iftheyremunerateexcessreserves,andontheirbalancesheetsastheyproceedtoactivelyshrinkthem.Inanextremeformoffiscaldominance,peoplemaybelievethatthecentralbankcouldtrytoinflateawaypartofthedebtorresorttooutrightmonetizationoffuturedeficits,whichinturnwouldleadinflationexpectationstoshootup.Thefiscaltheoryofthepricelevelhasreformulatedthesenotions,arguingthatpricelevelsadjustsothattherealvalueofgovernmentdebtequalsthepresentvalueoftaxeslessspending(Cochrane2023).Suchconcernsare,inprinciple,relevantforbothadvanced-andemergingmarketeconomies.Inmanyemergingmarketeconomies,theseissuesusedtobeverymuchattheforefront

untiltheearly2000s.Forexample,Celasun,Gelos,andPrati(2004)foundthatfiscalvariableswereanimportantdeterminant

ofinflationexpectationsinmajoremergingmarketeconomies.Fiscalconsolidation,theadoptionofmorecrediblemacroeconomicandmonetaryframeworks,andthegrantingof

autonomytocentralbanks

lessenedtheseconcernsintheseeconomiesoverthefollowingyears.

However,centralbankcredibility

remainslessestablishedinsomeemergingmarketeconomies,whichasagrouparegenerallystillmorevulnerabletoexternalshocks.Althoughsovereignforeigncurrencyexposureshavefallen,currencymismatchesintheprivatesectorremainwidespread(BIS,2019),posingvulnerabilitiesthatultimatelycanthreatenthefiscalpositionof

theseeconomies.Becauseofthishistoricalbackgroundandthestructuralspecificitiesofemergingmarketeconomies,inthisstudywedifferentiatebetweenadvanced-andemergingmarketeconomies.Severalotherstudieshaveempiricallyinvestigatedtherelationshipbetweenfiscalvariablesandinflationexpectationsinswithinflation-targetingcentralbanks(Cat?o&Terrones,2003;deMendon?a&Machado,2013;Celasun,Gelos,andPrati,2004;Coibion,Gorodnichenko,andWeber,2021).Thesepapersfindgenerallythattighteningfiscalbalancesarehelpfulinreducinginflationexpectationsandobservedinflation,withstrongereffectsin

economieswithhighandpersistentinflation.Thispapertakesasomewhatdifferentapproachandcoversamuchlongerandrecentperiod.INTERNATIONALMONETARYFUND5IMFWORKINGPAPERSDebtSurprisesandInflationExpectationsSpecifically,weinvestigatehow

sovereigndebtmayaffectinflationdynamicsbyexaminingitsimpactoninflationexpectations.Foridentificationpurposes,wefocusondebt

surprises.Wetraceouttheresponseofinflationexpectationstoanunexpectedincreaseinsovereigndebtusinglocalprojections.Sinceweareinterestedintheeffectsof

debtanddebtsustainability,toavoidcapturingdemand-sideeffectsrelatedtounexpectedfiscalexpansions,wefocuson

long-runexpectations(5yearsahead),undertheassumptionthatanyinflationaryeffectsofsuchdemandpressuresareshorter-lived.Tofurtherminimizedemandeffects(suchasinitialpricepressuresleadingtoa

de-anchoringofinflationexpectationsunrelatedtolonger-termdebtconcerns),wealsoconditiononexistingdebtlevels.Wefindthatinresponsetodebtsurprises,inflationexpectationsinemergingmarketeconomiesrisesignificantly,whereastheydonotinadvancedeconomies.Specifically,foremergingmarketeconomies,a10-percentage-pointincreaseinthegovernment-debt-to-GDPratiotriggersastatisticallysignificant20-basispointincreaseinlong-terminflationexpectationswithinthefirstyearaftertheshockandreachingapeakof70-basispointwithinthesecondyear.Theeffectismuchlargerinemergingmarketeconomiesthatdonotfeatureinflationtargetingregimescomparedtothosethatdo,becausetheseregimesaretypicallyassociatedwithsoundmonetaryframeworksandautonomouscentralbanks

thathelpcontainexpectationsoffiscaldominance.TheseresultsarebroadlyconsistentwithearlieronesofCat?oandTerrones(2005)whofindastrongpositiveassociationbetweendeficitsandinflationamonghigh-inflationanddevelopingcountries,butnotamonglow-inflationadvancedeconomies.However,inlinewiththenotionthatdebtsustainabilityconcernsarebehindourresults,theeffectoninflationexpectationsisnotpresentforemergingmarketeconomieswithlowsovereigndebtlevels.Thesefindingsareinlinewiththoseby

Kwon,McFarlane,andRobinson(2009),whoreportevidenceconsistentwiththenotionthattheriskofadebt-inflationtrapissignificantinhighlyindebtedeconomies.TheoreticalMechanismsWeareprimarilyinterestedintheextenttowhichgovernment

debtandconcernsregardingsolvencyofthesovereignmightaffectmonetarypolicy

andtheabilityofacentralbanktoachieveits

pricestabilityobjective.Alargetheoreticalliteratureestablishesdirectlinksbetween

governmentdebtandthepricelevel.Thestartingpointisfiscaldominance:thatacentralbanktightensitsmonetarypolicystanceinresponsetoariseininflationbylessthanitwouldotherwisebecauseofthelevelofgovernment

debt.Totheextentthatthegovernmentdebtisinlocalcurrency,

highgovernmentdebtmayleadcentralbankstolimitincreasesinthepolicyrateoutofconcernforthegovernment’ssolvency.Economicagents,inturn,anticipatethis

reaction,raisingtheirinflationexpectations.Intheextreme,firmsandhouseholdsmayfearthatthecentralbankmayresorttooutright

debtmonetization.Implicitinthesemechanismsisthatsome

portionof

governmentdebtisnotbackedbythegovernment’scurrentandINTERNATIONALMONETARYFUND6IMFWORKINGPAPERSDebtSurprisesandInflationExpectationsfutureprimarysurpluses.TheseargumentsarewellestablishedbySargentandWallace(1981)andinsubsequentforward-lookingmodelsofinflation(AiyagariandGertler,1985;Calvo,1988;Bohn,1988,amongothers).AlthoughSargentandWallace(1981)neverdiscussthefiscalstanceordebtsustainability,thekeypointisthatthefiscalpathispredeterminedandneedstogenerateseigniorage.Acorollaryoftheseargumentsisthatatightermonetarypolicymay

deliverlowerinflationnow(andinextremecasesnoteventhat)butsurelyhigherinflationinthefutureiffiscalpolicydominatesmonetarypolicy.Thisisbecause(sufficiently)forward-lookingagentsanticipatetheneedforloosermonetarypolicylatertoinflateawaythepublicdebt.AsimilarmechanismalbeitwithimportanttheoreticaldeparturesisdevelopedinthefiscaltheoryofthepricelevelorFTPL(Leeper1991,

Sims1994,Woodford1994,Cochrane2005,andBenigno2020).ThecentralmechanismintheFTPListhatpricesadjustsothattherealvalueofgovernmentdebtequalsthepresentvalueoftaxeslessspending(Cochrane,

2023).Morerecently,BianchiandMelosi(2022)developedatheoreticalframeworkwheremonetary-policyfiscalinteractionismoreacutewhenfiscalimbalancesarelargeandfiscalcredibilitywanes,makingitharderforthemonetaryauthoritytostabilizeinflationarounditsdesiredtarget.Intheirframework,afiscally-drivenriseintrendinflationdoesnotnecessarilystemfrom

alackofcredibilityofthecentralbank,butratherfromtheincompatibilitybetweentheobjectivesofthecentralbankandtheexpectedbehaviorofthefiscalauthorityincrediblystabilizingitsdebt.Adifferent,butrelatedmechanismdevelopedmorerecentlyin

Arellanoandothers(2020)arguesthatsovereigndefaultriskonexternaldefaultablegovernmentdebtaffectsinflationexpectations.Intheirmodel,byassumption,inflationishigh(andconsumptionlow)wheneverthesovereigndefaultbecauseproductivityislow.Specifically,theyassumethattotalfactorproductivitydependsonthegovernment’screditstanding.Takingexpectations,whendefaultriskis

high,expectedinflationishigh.Thus,anyshockthatincreasesthesovereign’sdefaultrisk—forinstance,aproductivityoratermsoftradeshock—raises(near-term)inflationexpectations.Inthispaper,

weidentifyshockstodebtthatareplausiblyorthogonaltoshocksthatmoveinflationandoutputandfindthattheseshocksarerelevantfortheconductofmonetarypolicyassummarizedbydislocationsof

long-termexpectationsinemergingmarketeconomies.Furthermore,weestablishthepresenceofnonlinearitiesbydemonstratinghighersensitivityofinflationexpectationsto

debtsurprisesin

high-debtandhigh-inflationemergingmarketeconomies.DataandEmpiricalMethodsEconometricApproachOurgoalofassessinghowdebtsustainabilityconcernsmightimpingeonmonetarypolicyanditsabilitytoanchorinflationexpectationsfacesvariousempiricalchallenges,mostlyrelatedtoendogeneity.First,intherelationshipbetweendebtlevels

andinflation,causalityprobablyflowsbothways.Forexample,policymakersmaytightenfiscal

policyinresponseto

ariseininflationexpectations,whichcouldresultinanegativecorrelationofchangesindebtandinflationexpectations(Celasun,Gelos,andPrati,2004).Ontheotherhand,fiscalpolicyactionswhichaffectINTERNATIONALMONETARYFUND7IMFWORKINGPAPERSDebtSurprisesandInflationExpectationsdebtmayalsodirectlyinfluenceinflationexpectationsthroughaggregatedemandeffects,whichwewouldliketoexclude.Second,theco-movementbetweendebtlevelsandinflationcouldbespuriousanddriveninsteadbyanomittedvariable.Particularlyinemergingmarketeconomies,adverseGDPshocksstemming,say,fromasuddenstopofcapitalflowsasaresultofchangesinglobalriskaversionoftencomealongwithexchangeratedepreciationsthatinduceariseininflationexpectations.

Insuchsituations,wewouldobserveasimultaneousincreaseinthedebt-to-GDPratioandinflationexpectationsthatisnot(necessarily)drivenbythemechanismswearetryingtocapture.Moreover,informationshocksthatmanifestaspositiveornegativeinterestrateshocks—reflectingacentralbankcommunicatingasurprisinglypositiveornegativeeconomicoutlook—couldcausehigher/lowernominaldebtalongwithhigherorlowerinflationexpectations(Nunesandothers,2022).Also,

othershockstoinflationexpectationsthathaveanimpactonthenominalinterestratebill(hence,ondebtsustainability)couldintroduceapositive(negative)biasforlocal(foreign)currencydenominatedinstruments(Celasunandothers,2004).Moreover,sinceweareinterestedinmechanismsrelatedtothegovernment’sintertemporalbudgetconstraint,as

discussedearlier,

totheextentpossible,wewouldliketoexcludeshort-terminflationarydemandeffectsstemmingfromfiscalexpansions.Toaddresstheseconcerns,wepursueathreefoldstrategy.First,we

useforecasterrorsfromvariousvintagesofInternationalMonetaryFund(IMF)WorldEconomicOutlook(WEO)publicationstoidentifyunanticipatedexogenousshocks(surprises)togovernmentdebt,whichbothmitigateendogeneityconcerns.Moreover,debtsurprisesmaycapturechangesindebtthatdonotnecessarilymoveone-for-onewiththebudgetdeficitandarelesslikelytobedrivenbychangesingovernment

expenditures(Singhandothers,2005).Forexample,suchdebtsurprisesmayincludetherecognitionofhiddencontingentliabilities.Second,wefocusonlong-terminflationexpectations(i.e.,5-yearaheadexpectations),whicharelesslikelytobeaffectedbyshort-termaggregatedemandeffects(assumingreasonableestimatesofaggregatepricestickiness).Third,tofurtherreducetheinfluenceofaggregatedemandeffects

andsimultaneity,wealsoconditionourestimationsonexistingdebtlevels;ifdebtsurprisesaffect

long-terminflationexpectationsmorethehigherpre-existingdebtlevelsare,thisis

unlikelytostemfromsuchdemandeffects

orothershocks.Thedebtshocksaredefinedasthedifferencebetweenthe

realizedannualgrowthrateofthedebttoGDPratioandthecorrespondingIMFforecast,whichispublishedinOctober.Inotherwords,theshockisunanticipatedchangeinthedebt-to-GDPratiothatisobservedinthelastquarterofeachyear.Specifically,thedebtsurprise,debt?????,forcountry

attime?

isdefinedas:?,?Debtrealized?,?GDPrealized?,??1Debtforecast?,?GDPrealized?,??1debtshock

=?,??wherethelowercasedebtvariablesdenotedebt-to-GDPratios.Asarguedintheliteraturethatusesthisapproach(e.g.,BlanchardandPerotti,2002;AuerbachandGorodnichenko,2012,2013;Abiadandothers,2016;FurceriandLi,2017),debtlevelsprojectedintheOctoberWEOpublicationsarelikelytoreflectallavailableinformationatthetimeofpublication.INTERNATIONALMONETARYFUND8IMFWORKINGPAPERSDebtSurprisesandInflationExpectationsAnydifferencebetweenrealizednominaldebtlevelsandtheOctoberWEOprojectionsthusaremorelikelytoreflectunanticipatedpolicychangesordebtrevaluations.Oneconcernwiththegovernment-debtforecasterrorsisthattheymaynotrepresenttruesurprises,becausetheymaybeforecastable.Toverifythesensitivityofourresultstosuchissues,weconsidertwoapproachesfromtheliterature.First,wefollowthemethodusedintheOctober2017

WEOandMagudandPienknagura(2022)byregressingthedebtforecasterrorsonforecasterrorsofinflationandrealGDPgrowthandusingtheresidualsfromthisregressionasourdebtshock.Theotherapproach,whichfollowsAuerbachandGorodnichenko(2013),

consistsofregressingthedebtforecasterrorsonlaggedmacroeconomicvariables—outputgap,debt,andprimarybalance—withwhichtheforecasterrorsmightbepredicted.Theresidualsfromthisregressionarethenusedasthedebtshock.Inbothapproaches,weincludeasetofcountryfixedeffects.

Theresultsandconclusionsdiscussedbelowarequantitativelyandqualitativelyrobusttothese

alternativeshocks.Aspreviouslydiscussed,government

debtshockscanaffectinflationexpectationsthroughmultiplechannels.Theimpactoftheseshocksonlong-termexpectationswilldependonthemonetarypolicyregime,thedegreeofcentralbankindependenceandotherinstitutionalfeatures,andtheanchoringofinflationexpectations.Thus,analyzingthesensitivityof

long-terminflationexpectationstounexpectedshiftsingovernmentdebthelpsshedlightonwhethernon-aggregatedemandmechanismsareatwork.WeusepanelregressionsandthelocalprojectionsmethodofJordà(2005)toestimatetheresponsesoflong-terminflationexpectationstotheidentifiedgovernmentdebtshocksoverdifferenthorizons.AsnotedbyAuerbachandGorodnichenko(2012),thelocalprojectionsapproacheasilyaccommodatesnonlinearitiesinimpulseresponsefunctionsandcross-countrycorrelationintheerrorterm.Thebaselinespecificationweestimateisgivenby:???(?)?

=

?

+

?

+

∑

?

??????,???+

∑

?

?

??

(

)?+

?

Γ+

??,?+?,

?

=

4,

10,

16,

22,

26,

and34,′(1)?,?+??????,??????=0?=1where?(?)?denotesthemeasureofinflationexpectationsforhorizonHforcountryi

attimet+l,?,?+????,???ud

,t

?s

aredebtsurprises,

′

Xisavectorofcontrols(tobeintroducedlaterasrobustness),

h?i

i??and

arecountryandtimefixedeffects,and??,?+?

isarandomdisturbance.Weincludelagsj

ofthethdebtshock

toreduceconcernsthatthedynamicsoftheimpactareduetopropertiesoftheshock(i.e.,thattheshockispersistentandtosomeextentpredictable)2

asopposedtotheoperatingmechanisms.Thecoefficients,??,oftheresponseofinflationexpectationstodebtsurprises02

Severalauthorshavepointedoutthatanalystforecasts,includingWEOforecasts,are

tosomeextentpredictable(e.g.,Celasunandothers2021forGDPforecasts).Debtforecasts,inparticular,showpositiveerrorsthatincreasewithdebt-to-GDPratios,andwhicharestatecontingent.Asnotedabove,forrobustnesswefilterthedebtsurprisesonotherendogenousvariableswithwhichdebtsurprisesmightbeforecastableandfindthatourresultsarequalitativelyandquantitativelysimilar.INTERNATIONALMONETARYFUND9IMFWORKINGPAPERSDebtSurprisesandInflationExpectationsspecificationareestimatedusingthewithinestimatorforeachprojectionperiod,withstandarderrorscomingfromDriscollandKraay’s(1998)heteroskedasticityandautocorrelation-consistentestimator.DataThesampleincludes41emergingmarketanddevelopingeconomies,7low-incomeanddevelopingcountries,and34advancedeconomiesbetween2000and2020.3

Wegrouptogetheremergingmarket-anddevelopingeconomies.Inflationexpectationsdataarecapturedusingsurvey-basedinflationforecastsfromprofessionalforecastersreportedbyConsensusEconomicsforfiveyearsahead.4

Weuseinflationexpectationsdataatasemi-annualfrequency

becauseformostofoursample,intheConsensusEconomicsforecasts,long-terminflationforecastsareonlycollectedatasemi-annualfrequency(inAprilandOctober).Ourchoiceofexpectationsfromprofessionalforecastersis

drivensolelybytheconsistentavailabilityacrosscountriesofinflationexpectationsdata.5

Dataongovernmentdebt,GDP,outputgap,andconsumerpriceinflationarefromtheIMF’sWEOdataset,whichcoversalleconomiesfrom1995to2019.Inconstructingourdebtshock,theforecastforgovernmentdebtandGDPforeachcountryinagivenyearistakenfromtheOctobervintageofWEOforecastsofthatyear.Therealizedvaluesofthesevariablesarethentakenfromthefollowingyear’sOctoberWEOdatabase.Table1.SummaryStatistics(Emerging)mean3.91(Advanced)mean2.04(Difference)bs.d.1.924.4918.277.33s.d.0.444.1444.621.71tInf.Expectations,5-yrDebtSurpriseDebttoGDPCPI,y-o-yObservations1.87***0.06-19.19***3.93***778(17.92)(0.19)(-7.59)(9.56)0.6537.915.620.5957.101.69352426WefollowFloresandothers(2023)incleaningthedatatominimizemeasurementerrors

andinconsistenciesacrossdatavintages.Figure1showstheavailabilityofdebtshocksdataforadvanced-andemergingmarketeconomies.

ThedebtforecastdatainWEOvintagesareavailableforasmallnumberofadvancedeconomiesbefore2002,whichimposesthestartingyearofthesampleforourempiricalanalysis.Figure2summarizesthedebtshocksforadvancedandemergingmarketeconomiesbypresentingtheinterquartilerange,themean,andmedianforecasterrorsbygroupofeconomies.ThemeanandmediandebttoGDPforecasterrorsaresimilaracrossthetwogroups.Themedianforecasterroris0.15foremergingmarketeconomies,and0.05foradvanced3

TableA.1inAnnexIcontainsthelistofeconomiesincludedinoursample.4

AnnexIIpresentsresultswhenweusethree-yearaheadinflationexpectations.Ourmainconclusionsareunalteredwhenconsideringashorterforecasthorizon.