Pitfalls of estimating the marginal likelihood using the modified harmonic mean

Joshua C.C. Chan; Angelia L. Grant

doi:10.1016/j.econlet.2015.03.036

Pitfalls of estimating the marginal likelihood using the modified harmonic mean

Joshua C.C. Chan^*, Angelia L. Grant

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

19 Citations (Scopus)

Abstract

The modified harmonic mean is widely used for estimating the marginal likelihood. We investigate the empirical performance of two versions of this estimator: one based on the observed-data likelihood and the other on the complete-data likelihood. Through an empirical example using US and UK inflation, we show that the version based on the complete-data likelihood has a substantial bias and tends to select the wrong model, whereas the version based on the observed-data likelihood works well.

Original language	English
Pages (from-to)	29-33
Number of pages	5
Journal	Economics Letters
Volume	131
DOIs	https://doi.org/10.1016/j.econlet.2015.03.036
Publication status	Published - 1 Jun 2015

Access to Document

10.1016/j.econlet.2015.03.036

Cite this

@article{c4e9eda59d7847f1abb2957949d6585c,

title = "Pitfalls of estimating the marginal likelihood using the modified harmonic mean",

abstract = "The modified harmonic mean is widely used for estimating the marginal likelihood. We investigate the empirical performance of two versions of this estimator: one based on the observed-data likelihood and the other on the complete-data likelihood. Through an empirical example using US and UK inflation, we show that the version based on the complete-data likelihood has a substantial bias and tends to select the wrong model, whereas the version based on the observed-data likelihood works well.",

keywords = "Bayesian model comparison, Inflation, State space, Unobserved components",

author = "Chan, \{Joshua C.C.\} and Grant, \{Angelia L.\}",

note = "Publisher Copyright: {\textcopyright} 2015 Elsevier B.V.",

year = "2015",

month = jun,

day = "1",

doi = "10.1016/j.econlet.2015.03.036",

language = "English",

volume = "131",

pages = "29--33",

journal = "Economics Letters",

issn = "0165-1765",

publisher = "Elsevier B.V.",

}

TY - JOUR

T1 - Pitfalls of estimating the marginal likelihood using the modified harmonic mean

AU - Chan, Joshua C.C.

AU - Grant, Angelia L.

PY - 2015/6/1

Y1 - 2015/6/1

N2 - The modified harmonic mean is widely used for estimating the marginal likelihood. We investigate the empirical performance of two versions of this estimator: one based on the observed-data likelihood and the other on the complete-data likelihood. Through an empirical example using US and UK inflation, we show that the version based on the complete-data likelihood has a substantial bias and tends to select the wrong model, whereas the version based on the observed-data likelihood works well.

AB - The modified harmonic mean is widely used for estimating the marginal likelihood. We investigate the empirical performance of two versions of this estimator: one based on the observed-data likelihood and the other on the complete-data likelihood. Through an empirical example using US and UK inflation, we show that the version based on the complete-data likelihood has a substantial bias and tends to select the wrong model, whereas the version based on the observed-data likelihood works well.

KW - Bayesian model comparison

KW - Inflation

KW - State space

KW - Unobserved components

UR - http://www.scopus.com/inward/record.url?scp=84930209027&partnerID=8YFLogxK

U2 - 10.1016/j.econlet.2015.03.036

DO - 10.1016/j.econlet.2015.03.036

M3 - Article

SN - 0165-1765

VL - 131

SP - 29

EP - 33

JO - Economics Letters

JF - Economics Letters

ER -

Pitfalls of estimating the marginal likelihood using the modified harmonic mean

Abstract

Access to Document

Other files and links

Fingerprint

Cite this