Benchmark priors for Bayesian model averaging

In contrast to a posterior analysis given a particular sampling model, posterior model probabilities in the context of model uncertainty are typically rather sensitive to the specification of the prior. In particular, 'diffuse' priors on model-specific parameters can lead to quite unexpect...

ver descrição completa

Detalhes bibliográficos
Autores: Fernández-Llana, Carmen, Ley, Eduardo, Steel, Mark F.J.
Formato: artículo
Estado:Versión aceptada para publicación
Fecha de publicación:2000
País:España
Recursos:Consejo Superior de Investigaciones Científicas (CSIC)
Repositorio:DIGITAL.CSIC. Repositorio Institucional del CSIC
OAI Identifier:oai:digital.csic.es:10261/341960
Acesso em linha:http://hdl.handle.net/10261/341960
https://api.elsevier.com/content/abstract/scopus_id/18044404766
Access Level:acceso abierto
Palavra-chave:Bayes factors
Markov chain Monte Carlo
Posterior odds
Prior elicitation
id ES_0b5aa2ed7e832c0ac97cbfea5a3c7003
oai_identifier_str oai:digital.csic.es:10261/341960
network_acronym_str ES
network_name_str España
repository_id_str
spelling Benchmark priors for Bayesian model averagingFernández-Llana, CarmenLey, EduardoSteel, Mark F.J.Bayes factorsMarkov chain Monte CarloPosterior oddsPrior elicitationIn contrast to a posterior analysis given a particular sampling model, posterior model probabilities in the context of model uncertainty are typically rather sensitive to the specification of the prior. In particular, 'diffuse' priors on model-specific parameters can lead to quite unexpected consequences. Here we focus on the practically relevant situation where we need to entertain a (large) number of sampling models and we have (or wish to use) little or no subjective prior information. We aim at providing an 'automatic' or 'benchmark' prior structure that can be used in such cases. We focus on the normal linear regression model with uncertainty in the choice of regressors. We propose a partly non-informative prior structure related to a natural conjugate g-prior specification, where the amount of subjective information requested from the user is limited to the choice of a single scalar hyperparameter g0j. The consequences of different choices for g0j are examined. We investigate theoretical properties, such as consistency of the implied Bayesian procedure. Links with classical information criteria are provided. More importantly, we examine the finite sample implications of several choices of g0j in a simulation study. The use of the MC3 algorithm of Madigan and York (Int. Stat. Rev. 63 (1995) 215), combined with efficient coding in Fortran, makes it feasible to conduct large simulations. In addition to posterior criteria, we shall also compare the predictive performance of different priors. A classic example concerning the economics of crime will also be provided and contrasted with results in the literature. The main findings of the paper will lead us to propose a 'benchmark' prior specification in a linear regression context with model uncertainty. © 2001 Elsevier Science S.A. All rights reserved.Carmen Fernández gratefully acknowledges financial support from a Training and Mobility of Researchers grant awarded by the European Commission (ERBFMBICT # 961021). Carmen Fernández and Mark Steel were affiliated to CentER and the Department of Econometrics, Tilburg University, The Netherlands, and Eduardo Ley was at FEDEA, Madrid, Spain during the early stages of the work on this paper. Some of this research was done when Carmen Fernández was at the Department of Mathematics, University of Bristol, and Mark Steel at the Department of Economics, University of Edinburgh.Peer reviewedElsevierEuropean CommissionTilburg UniversityUniversity of Bristol202420242000info:eu-repo/semantics/articlehttp://purl.org/coar/resource_type/c_6501Postprintinfo:eu-repo/semantics/acceptedVersionhttp://hdl.handle.net/10261/341960https://api.elsevier.com/content/abstract/scopus_id/18044404766reponame:DIGITAL.CSIC. Repositorio Institucional del CSICinstname:Consejo Superior de Investigaciones Científicas (CSIC)InglésJournal of Econometricshttps://doi.org/10.1016/S0304-4076(00)00076-2Noinfo:eu-repo/semantics/openAccessoai:digital.csic.es:10261/3419602026-05-22T06:33:51Z
dc.title.none.fl_str_mv Benchmark priors for Bayesian model averaging
title Benchmark priors for Bayesian model averaging
spellingShingle Benchmark priors for Bayesian model averaging
Fernández-Llana, Carmen
Bayes factors
Markov chain Monte Carlo
Posterior odds
Prior elicitation
title_short Benchmark priors for Bayesian model averaging
title_full Benchmark priors for Bayesian model averaging
title_fullStr Benchmark priors for Bayesian model averaging
title_full_unstemmed Benchmark priors for Bayesian model averaging
title_sort Benchmark priors for Bayesian model averaging
dc.creator.none.fl_str_mv Fernández-Llana, Carmen
Ley, Eduardo
Steel, Mark F.J.
author Fernández-Llana, Carmen
author_facet Fernández-Llana, Carmen
Ley, Eduardo
Steel, Mark F.J.
author_role author
author2 Ley, Eduardo
Steel, Mark F.J.
author2_role author
author
dc.contributor.none.fl_str_mv European Commission
Tilburg University
University of Bristol
dc.subject.none.fl_str_mv Bayes factors
Markov chain Monte Carlo
Posterior odds
Prior elicitation
topic Bayes factors
Markov chain Monte Carlo
Posterior odds
Prior elicitation
description In contrast to a posterior analysis given a particular sampling model, posterior model probabilities in the context of model uncertainty are typically rather sensitive to the specification of the prior. In particular, 'diffuse' priors on model-specific parameters can lead to quite unexpected consequences. Here we focus on the practically relevant situation where we need to entertain a (large) number of sampling models and we have (or wish to use) little or no subjective prior information. We aim at providing an 'automatic' or 'benchmark' prior structure that can be used in such cases. We focus on the normal linear regression model with uncertainty in the choice of regressors. We propose a partly non-informative prior structure related to a natural conjugate g-prior specification, where the amount of subjective information requested from the user is limited to the choice of a single scalar hyperparameter g0j. The consequences of different choices for g0j are examined. We investigate theoretical properties, such as consistency of the implied Bayesian procedure. Links with classical information criteria are provided. More importantly, we examine the finite sample implications of several choices of g0j in a simulation study. The use of the MC3 algorithm of Madigan and York (Int. Stat. Rev. 63 (1995) 215), combined with efficient coding in Fortran, makes it feasible to conduct large simulations. In addition to posterior criteria, we shall also compare the predictive performance of different priors. A classic example concerning the economics of crime will also be provided and contrasted with results in the literature. The main findings of the paper will lead us to propose a 'benchmark' prior specification in a linear regression context with model uncertainty. © 2001 Elsevier Science S.A. All rights reserved.
publishDate 2000
dc.date.none.fl_str_mv 2000
2024
2024
dc.type.none.fl_str_mv info:eu-repo/semantics/article
http://purl.org/coar/resource_type/c_6501
Postprint
info:eu-repo/semantics/acceptedVersion
format article
status_str acceptedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/10261/341960
https://api.elsevier.com/content/abstract/scopus_id/18044404766
url http://hdl.handle.net/10261/341960
https://api.elsevier.com/content/abstract/scopus_id/18044404766
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv Journal of Econometrics
https://doi.org/10.1016/S0304-4076(00)00076-2
No
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.publisher.none.fl_str_mv Elsevier
publisher.none.fl_str_mv Elsevier
dc.source.none.fl_str_mv reponame:DIGITAL.CSIC. Repositorio Institucional del CSIC
instname:Consejo Superior de Investigaciones Científicas (CSIC)
instname_str Consejo Superior de Investigaciones Científicas (CSIC)
reponame_str DIGITAL.CSIC. Repositorio Institucional del CSIC
collection DIGITAL.CSIC. Repositorio Institucional del CSIC
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869403216085516288
score 15.812429