Record Nr.

UNINA9910812315003321

Autore

Feldkircher Martin

Titolo

Benchmark priors revisited : on adaptive shrinkage and the supermodel effect in Bayesian Model averaging / / prepared by Martin Feldkircher and Stefan Zeugner

Pubbl/distr/stampa


[Washington, DC], : International Monetary Fund, Finance Dept., c2009

ISBN

1-4623-4466-6

1-4518-7349-2

9786612844096

1-4527-6923-0

1-282-84409-1

Edizione

[1st ed.]

Descrizione fisica

39 p. : col. ill

Collana

IMF working paper ; ; WP/09/202

Altri autori (Persone)

ZeugnerStefan

Disciplina

332.015195

Soggetti

Bayesian statistical decision theory

Economic development - Mathematical models

Lingua di pubblicazione

Inglese

Formato

Materiale a stampa

Livello bibliografico

Monografia

Note generali

"September 2009."

Nota di bibliografia

Includes bibliographical references.

Nota di contenuto

Cover Page -- Title Page -- Copyright Page -- Contents -- I Introduction -- II Bayesian Model Averaging under Zellner's g Prior -- II. 1 Popular Settings for Zellner's g -- III The Hyper-g Prior: A Beta Prior on the Shrinkage Factor -- IV A Simulation Study -- V Growth Determinants Revisited -- VI Concluding Remarks -- A Technical Appendix -- A. 1 Consistency of the Hyper-g Prior -- A. 2 Relationship between Hyper-g Prior and EBL -- A. 3 The Shrinkage Factor and Goodness-of-Fit -- A. 4 The Posterior Predictive Distribution and the Hyper-g Prior -- A. 5 The Beta-binomial Prior over the Model Space -- A. 6 Charts and Tables -- References -- Footnotes.

Sommario/riassunto

Default prior choices fixing Zellner's g are predominant in the Bayesian Model Averaging literature, but tend to concentrate posterior mass on a tiny set of models. The paper demonstrates this supermodel effect and proposes to address it by a hyper-g prior, whose data-dependent shrinkage adapts posterior model distributions to data quality. Analytically, existing work on the hyper-g-prior is complemented by posterior expressions essential to fully Bayesian analysis and to sound

numerical implementation. A simulation experiment illustrates the implications for posterior inference. Furthermore, an application to determinants of economic growth identifies several covariates whose robustness differs considerably from previous results.