Weighted Median Inflation Around the World: A Measure of Core Inflation

WeightedMedianInflationAroundtheWorld:AMeasureofCoreInflation LaurenceBall,CarlosCarvalho,ChristopherEvansandLucaAntonioRicci WP/23/44 IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate. TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement. 2023 FEB ©2023InternationalMonetaryFundWP/23/44 IMFWorkingPaper AfricanDepartment WeightedMedianInflationAroundtheWorld:AMeasureofCoreInflation PreparedbyLaurenceBall,CarlosCarvalho,ChristopherEvansandLucaAntonioRicci* AuthorizedfordistributionbyLucaAntonioRicciFebruary2023 IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement. ABSTRACT:Thestandardmeasureofcoreorunderlyinginflationistheinflationrateexcludingfoodandenergyprices.Thispaperconstructsanalternativemeasure,theweightedmedianinflationrate,for38advancedandemergingeconomiesusingsubclassleveldisaggretionoftheCPIover1990-2021,andcomparesthepropertiesofthismeasuretothoseofstandardcore.Forquarterlydata,wefindthattheweightedmedianislessvolatilethanstandardcore,morecloselyrelatedtoeconomicslack,andmorecloselyrelatedtoheadlineinflationoverthenextyear.Theweightedmedianalsohasadrawback:inmostcountries,ithasaloweraveragelevelthanheadlineinflation.Wethereforealsoconsiderameasureofcoreinflationthateliminatesthisbias,whichisbasedonthepercentileofsectoralinflationratesthatmatchesthesampleaverageofheadlineCPIinflation. RECOMMENDEDCITATION:Ball,Laurence,CarlosCarvalho,ChristopherEvansandLucaAntonioRicci,2023,“WeightedMedianInflationAroundtheWorld:AMeasureofCoreInflation”,IMF,WorkingPaperWP/23/44 JELClassificationNumbers: C43,E31,E52 Keywords: coreinflation,medianinflation,Phillipscurve. Author’sE-MailAddress: lball@jhu.edu,cvianac@econ.puc-rio.br,cevans@imf.org,lricci@imf.org *TheauthorswouldliketothankPedroHenriquedaSilvaCastroandIvanBurgaraforinvaluablediscussionsandsupportaswellasparticipantsatvariousIMFseminarsfortheirhelpfulfeedback.TheviewsexpressedherearethoseoftheauthorsanddonotnecessarilyrepresentthoseoftheIMF,itsManagementandExecutiveBoard,orIMFpolicy. Contents I.Introduction4 II.DataandMethodology6 III.Results8 A.Volatility8 B.AverageInflation9 C.PhillipsCurveEstimation10 D.P*,AnUnbiasedCoreMeasure13 E.P*:ResultsforVolatilityandBias14 F.P*:ResultsforPhillipsCurves16 G.MedianandP*:CapturingOne-Year-AheadHeadlineInflation17 H.2021-22InflationEvent18 IV.Conclusions19 References20 Appendixes22 A.MonthlyResults22 A.1ResultsonVolatility22 A.2ResultsonMean23 A.4ResultsonVolatilitywithP*24 A.5ResultsonBiastoHeadlinewithP*25 A.6ResultsonCapturingOne-Year-AheadHeadlineInflationwithP*26 B.AdditionalQuarterlyResults26 C.Investigatingthebias31 D.PanelPhillipsCurveRegressionTables31 FIGURES Figure1.Quarterly,Volatility9 Figure2.Quarterly,Mean10 Figure3.Quarterly,BiastoHeadline10 Figure4.PhillipsCurve,BaselineSpecification,R211 Figure5.PhillipsCurve,BaselineSpecification,R2,ShowingOnlyCaseswithPositiveandSignificantCoefficientsonOutput12 Figure6.PanelPhillipsCurve,R212 Figure7.PercentileAssociatedwithP*14 Figure8.QuarterlywithP*,Volatility15 Figure9.QuarterlywithP*,BiastoHeadline15 Figure10.PhillipsCurve,BaselineSpecificationwithP*,R216 Figure11.PanelPhillipsCurvewithP*,R217 Figure12.Quarterly,RMSEofCoreMeasures/RMSEofHeadlineInflation18 Figure13.12-Month,SelectedCountryPercentiles19 Figure14.Monthly,Volatility22 Figure15.Monthly,VolatilityofDifferenceinInflation23 Figure16.Monthly,Mean23 Figure17.Monthly,BiastoHeadline24 Figure18.MonthlywithP*,Volatility25 Figure19.MonthlywithP*,BiastoHeadline27 Figure20.Monthly,RMSEofCoreMeasures/RMSEofHeadlineInflation26 Figure21.Quarterly,VolatilityofDifferenceinInflation27 Figure22.Quarterly,BiasversusMean27 Figure23.PhillipsCurve,HybridSpecification,R228 Figure24.PhillipsCurve,HybridSpecification,R2ShowingOnlyResultsThatIncludeSignificantCoefficients ............................................................................................................................................................................29 Figure25.QuarterlywithP*,VolatilityofDifferenceInflation29 Figure26.PhillipsCurve,HybridSpecificationwithP*,R230 Figure27.12-Month,RandomWalk,RMSECoreMeasures/RMSEofHeadlineInflation30 Figure28.MonthlyandQuarterly,BiasVersusShareofZeros31 Figure29.PhillipsCurve,BaselineSpecification,DifferenceinR2withHeadline34 Figure30.PhillipsCurve,BaselineSpecification,R2andCoefficientontheOutputGap34 TABLES Table1CountrySample6 Table2PhillipsCurve,BaselineSpecification,R232 Table3PhillipsCurve,BaselineSpecification,CoefficientontheOutputGap33 I.Introduction Economistshavelongsoughtagoodmeasureoftheunderlyingorcoreinflationrate,onethatfiltersouttransitoryfluctuationsinheadlineinflationarisingfromsectoralshocks.Gordon(1975)proposedasimplemeasureofcore:theinflation