Weighted-average least squares estimation of generalized linear models

Giuseppe De Luca, Jan R. Magnus, Franco Peracchi

Research output: Contribution to JournalArticleAcademicpeer-review

Abstract

The weighted-average least squares (WALS) approach, introduced by Magnus et al. (2010) in the context of Gaussian linear models, has been shown to enjoy important advantages over other strictly Bayesian and strictly frequentist model-averaging estimators when accounting for problems of uncertainty in the choice of the regressors. In this paper we extend the WALS approach to deal with uncertainty about the specification of the linear predictor in the wider class of generalized linear models (GLMs). We study the large-sample properties of the WALS estimator for GLMs under a local misspecification framework, and the finite-sample properties of this estimator by a Monte Carlo experiment the design of which is based on a real empirical analysis of attrition in the first two waves of the Survey of Health, Aging and Retirement in Europe (SHARE).

Original languageEnglish
Pages (from-to)1-17
Number of pages17
JournalJournal of Econometrics
Volume204
Issue number1
DOIs
Publication statusPublished - 1 May 2018

Fingerprint

Estimator
Least squares
Uncertainty
Generalized linear model
Monte Carlo experiment
Finite sample properties
Health
Retirement
Misspecification
Predictors
Empirical analysis
Attrition
Model averaging
Least squares estimator

Keywords

  • Attrition
  • Generalized linear models
  • Model averaging
  • Monte Carlo
  • WALS

Cite this

De Luca, Giuseppe ; Magnus, Jan R. ; Peracchi, Franco. / Weighted-average least squares estimation of generalized linear models. In: Journal of Econometrics. 2018 ; Vol. 204, No. 1. pp. 1-17.
@article{7cb132395bf34c6cab0e138ee35e166e,
title = "Weighted-average least squares estimation of generalized linear models",
abstract = "The weighted-average least squares (WALS) approach, introduced by Magnus et al. (2010) in the context of Gaussian linear models, has been shown to enjoy important advantages over other strictly Bayesian and strictly frequentist model-averaging estimators when accounting for problems of uncertainty in the choice of the regressors. In this paper we extend the WALS approach to deal with uncertainty about the specification of the linear predictor in the wider class of generalized linear models (GLMs). We study the large-sample properties of the WALS estimator for GLMs under a local misspecification framework, and the finite-sample properties of this estimator by a Monte Carlo experiment the design of which is based on a real empirical analysis of attrition in the first two waves of the Survey of Health, Aging and Retirement in Europe (SHARE).",
keywords = "Attrition, Generalized linear models, Model averaging, Monte Carlo, WALS",
author = "{De Luca}, Giuseppe and Magnus, {Jan R.} and Franco Peracchi",
year = "2018",
month = "5",
day = "1",
doi = "10.1016/j.jeconom.2017.12.007",
language = "English",
volume = "204",
pages = "1--17",
journal = "Journal of Econometrics",
issn = "0304-4076",
publisher = "Elsevier BV",
number = "1",

}

Weighted-average least squares estimation of generalized linear models. / De Luca, Giuseppe; Magnus, Jan R.; Peracchi, Franco.

In: Journal of Econometrics, Vol. 204, No. 1, 01.05.2018, p. 1-17.

Research output: Contribution to JournalArticleAcademicpeer-review

TY - JOUR

T1 - Weighted-average least squares estimation of generalized linear models

AU - De Luca, Giuseppe

AU - Magnus, Jan R.

AU - Peracchi, Franco

PY - 2018/5/1

Y1 - 2018/5/1

N2 - The weighted-average least squares (WALS) approach, introduced by Magnus et al. (2010) in the context of Gaussian linear models, has been shown to enjoy important advantages over other strictly Bayesian and strictly frequentist model-averaging estimators when accounting for problems of uncertainty in the choice of the regressors. In this paper we extend the WALS approach to deal with uncertainty about the specification of the linear predictor in the wider class of generalized linear models (GLMs). We study the large-sample properties of the WALS estimator for GLMs under a local misspecification framework, and the finite-sample properties of this estimator by a Monte Carlo experiment the design of which is based on a real empirical analysis of attrition in the first two waves of the Survey of Health, Aging and Retirement in Europe (SHARE).

AB - The weighted-average least squares (WALS) approach, introduced by Magnus et al. (2010) in the context of Gaussian linear models, has been shown to enjoy important advantages over other strictly Bayesian and strictly frequentist model-averaging estimators when accounting for problems of uncertainty in the choice of the regressors. In this paper we extend the WALS approach to deal with uncertainty about the specification of the linear predictor in the wider class of generalized linear models (GLMs). We study the large-sample properties of the WALS estimator for GLMs under a local misspecification framework, and the finite-sample properties of this estimator by a Monte Carlo experiment the design of which is based on a real empirical analysis of attrition in the first two waves of the Survey of Health, Aging and Retirement in Europe (SHARE).

KW - Attrition

KW - Generalized linear models

KW - Model averaging

KW - Monte Carlo

KW - WALS

UR - http://www.scopus.com/inward/record.url?scp=85041566223&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85041566223&partnerID=8YFLogxK

U2 - 10.1016/j.jeconom.2017.12.007

DO - 10.1016/j.jeconom.2017.12.007

M3 - Article

VL - 204

SP - 1

EP - 17

JO - Journal of Econometrics

JF - Journal of Econometrics

SN - 0304-4076

IS - 1

ER -