欧洲央行-一种稳健的倾斜方法：参数相对熵.pdfVIP

WorkingPaperSeries

CarlosMontes-Galdón,JoanParedes,Arobustapproachtotilting:

EliasWolf

parametricrelativeentropy

No3200

Disclaimer:ThispapershouldnotbereportedasrepresentingtheviewsoftheEuropeanCentralBank

(ECB).TheviewsexpressedarethoseoftheauthorsanddonotnecessarilyreflectthoseoftheECB.

Abstract

Weintroduceanovelmethodology,”parametrictilting,”forincorporating

externalinformationintoeconometricmodel-baseddensityforecasts.Unlike

traditionalentropictilting,whichcangenerateunrealisticorunstabledis-

tributionsundercertainconditions,parametrictiltingensuresmorereliable

andnumericallystableresults.Ourapproachleveragestheflexibilityofthe

skew-Tdistribution,whichcaptureskeymomentsofmacroeconomictime

series,andminimizestheKullback-Leiblerdivergencebetweenthetargetand

model-baseddistributions.Thismethodovercomeslimitationsofentropic

tilting,suchasmultimodalordegeneratedistributions,providingarobust

alternativeforpolicymakersandresearchersaimingtointegrateexternal

viewsintoprobabilisticforecastingframeworks.

JELCLASSIFICATIONSYSTEM:C14,C53,E52

KEYWORDS:Forecasting,Kullback-LeiblerInformationCriterion,Entropic

Tilting

ECBWorkingPaperSeriesNo32001

Non-TechnicalSummary

Inthispaper,wepresentanewandrobustapproachtointegrateexternalinfor-

mationintomodel-basedpredictions.Economistsandpolicymakerso