Bayesian and Frequentist Inference in Partially Identified Models

Hyungsik Roger Moon, Frank Schorfheide

Research output: Contribution to journalArticle

46 Citations (Scopus)

Abstract

A large-sample approximation of the posterior distribution of partially identified structural parameters is derived for models that can be indexed by an identifiable finite-dimensional reduced-form parameter vector. It is used to analyze the differences between Bayesian credible sets and frequentist confidence sets. We define a plug-in estimator of the identified set and show that asymptotically Bayesian highest-posterior-density sets exclude parts of the estimated identified set, whereas it is well known that frequentist confidence sets extend beyond the boundaries of the estimated identified set. We recommend reporting estimates of the identified set and information about the conditional prior along with Bayesian credible sets. A numerical illustration for a two-player entry game is provided.

Original languageEnglish
Pages (from-to)755-782
Number of pages28
JournalEconometrica
Volume80
Issue number2
DOIs
Publication statusPublished - 2012 Mar 1

Fingerprint

Inference
Confidence set
Reduced form
Estimator
Posterior distribution
Structural parameters
Approximation

All Science Journal Classification (ASJC) codes

  • Economics and Econometrics

Cite this

Moon, Hyungsik Roger ; Schorfheide, Frank. / Bayesian and Frequentist Inference in Partially Identified Models. In: Econometrica. 2012 ; Vol. 80, No. 2. pp. 755-782.
@article{2ca153d6eb2c42e6908d19595cc7457b,
title = "Bayesian and Frequentist Inference in Partially Identified Models",
abstract = "A large-sample approximation of the posterior distribution of partially identified structural parameters is derived for models that can be indexed by an identifiable finite-dimensional reduced-form parameter vector. It is used to analyze the differences between Bayesian credible sets and frequentist confidence sets. We define a plug-in estimator of the identified set and show that asymptotically Bayesian highest-posterior-density sets exclude parts of the estimated identified set, whereas it is well known that frequentist confidence sets extend beyond the boundaries of the estimated identified set. We recommend reporting estimates of the identified set and information about the conditional prior along with Bayesian credible sets. A numerical illustration for a two-player entry game is provided.",
author = "Moon, {Hyungsik Roger} and Frank Schorfheide",
year = "2012",
month = "3",
day = "1",
doi = "10.3982/ECTA8360",
language = "English",
volume = "80",
pages = "755--782",
journal = "Econometrica",
issn = "0012-9682",
publisher = "Wiley-Blackwell",
number = "2",

}

Bayesian and Frequentist Inference in Partially Identified Models. / Moon, Hyungsik Roger; Schorfheide, Frank.

In: Econometrica, Vol. 80, No. 2, 01.03.2012, p. 755-782.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Bayesian and Frequentist Inference in Partially Identified Models

AU - Moon, Hyungsik Roger

AU - Schorfheide, Frank

PY - 2012/3/1

Y1 - 2012/3/1

N2 - A large-sample approximation of the posterior distribution of partially identified structural parameters is derived for models that can be indexed by an identifiable finite-dimensional reduced-form parameter vector. It is used to analyze the differences between Bayesian credible sets and frequentist confidence sets. We define a plug-in estimator of the identified set and show that asymptotically Bayesian highest-posterior-density sets exclude parts of the estimated identified set, whereas it is well known that frequentist confidence sets extend beyond the boundaries of the estimated identified set. We recommend reporting estimates of the identified set and information about the conditional prior along with Bayesian credible sets. A numerical illustration for a two-player entry game is provided.

AB - A large-sample approximation of the posterior distribution of partially identified structural parameters is derived for models that can be indexed by an identifiable finite-dimensional reduced-form parameter vector. It is used to analyze the differences between Bayesian credible sets and frequentist confidence sets. We define a plug-in estimator of the identified set and show that asymptotically Bayesian highest-posterior-density sets exclude parts of the estimated identified set, whereas it is well known that frequentist confidence sets extend beyond the boundaries of the estimated identified set. We recommend reporting estimates of the identified set and information about the conditional prior along with Bayesian credible sets. A numerical illustration for a two-player entry game is provided.

UR - http://www.scopus.com/inward/record.url?scp=84858592506&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84858592506&partnerID=8YFLogxK

U2 - 10.3982/ECTA8360

DO - 10.3982/ECTA8360

M3 - Article

AN - SCOPUS:84858592506

VL - 80

SP - 755

EP - 782

JO - Econometrica

JF - Econometrica

SN - 0012-9682

IS - 2

ER -