Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/18876 
Full metadata record
Appears in Collections:
DC FieldValueLanguage
dc.contributor.authorTimmermann, Allanen
dc.contributor.authorPettenuzzo, Davideen
dc.contributor.authorPesaran, Mohammad Hashemen
dc.date.accessioned2009-01-28T15:53:30Z-
dc.date.available2009-01-28T15:53:30Z-
dc.date.issued2004-
dc.identifier.urihttp://hdl.handle.net/10419/18876-
dc.description.abstractThis paper provides a novel approach to forecasting time series subject to discrete structuralbreaks. We propose a Bayesian estimation and prediction procedure that allows for thepossibility of new breaks over the forecast horizon, taking account of the size and duration ofpast breaks (if any) by means of a hierarchical hidden Markov chain model. Predictions areformed by integrating over the hyper parameters from the meta distributions that characterizethe stochastic break point process. In an application to US Treasury bill rates, we find that themethod leads to better out-of-sample forecasts than alternative methods that ignore breaks,particularly at long horizons.en
dc.language.isoengen
dc.publisher|aCenter for Economic Studies and ifo Institute (CESifo) |cMunichen
dc.relation.ispartofseries|aCESifo Working Paper |x1237en
dc.subject.jelC53en
dc.subject.jelC11en
dc.subject.jelC15en
dc.subject.ddc330en
dc.subject.keywordstructural breaksen
dc.subject.keywordforecastingen
dc.subject.keywordhierarchical hidden Markov chain modelen
dc.subject.keywordBayesian model averagingen
dc.subject.stwPrognoseverfahrenen
dc.subject.stwZeitreihenanalyseen
dc.subject.stwStrukturbruchen
dc.subject.stwTheorieen
dc.titleForecasting time series subject to multiple structural breaks-
dc.typeWorking Paperen
dc.identifier.ppn856717800en
dc.rightshttp://www.econstor.eu/dspace/Nutzungsbedingungenen

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.