Bayesian Factor Model Shrinkage for Linear IV Regression With Many Instruments

Paul Hahn, Jingyu He, Hedibert Lopes

Research output: Contribution to journalArticle

Abstract

A Bayesian approach for the many instruments problem in linear instrumental variable models is presented. The new approach has two components. First, a slice sampler is developed, which leverages a decomposition of the likelihood function that is a Bayesian analogue to two-stage least squares. The new sampler permits nonconjugate shrinkage priors to be implemented easily and efficiently. The new computational approach permits a Bayesian analysis of problems that were previously infeasible due to computational demands that scaled poorly in the number of regressors. Second, a new predictor-dependent shrinkage prior is developed specifically for the many instruments setting. The prior is constructed based on a factor model decomposition of the matrix of observed instruments, allowing many instruments to be incorporated into the analysis in a robust way. Features of the new method are illustrated via a simulation study and three empirical examples.

Original languageEnglish (US)
Pages (from-to)278-287
Number of pages10
JournalJournal of Business and Economic Statistics
Volume36
Issue number2
DOIs
StatePublished - Apr 3 2018
Externally publishedYes

Fingerprint

Factor Models
Bayesian Model
Shrinkage
Regression
regression
Two-stage Least Squares
Decompose
Instrumental Variables
Bayesian Analysis
Likelihood Function
Bayesian Approach
Leverage
Slice
Predictors
Simulation Study
Analogue
simulation
Dependent
Decomposition
Model

Keywords

  • Bayesian econometrics
  • Horseshoe prior
  • Instrumental variables
  • Slice sampler

ASJC Scopus subject areas

  • Statistics and Probability
  • Social Sciences (miscellaneous)
  • Economics and Econometrics
  • Statistics, Probability and Uncertainty

Cite this

Bayesian Factor Model Shrinkage for Linear IV Regression With Many Instruments. / Hahn, Paul; He, Jingyu; Lopes, Hedibert.

In: Journal of Business and Economic Statistics, Vol. 36, No. 2, 03.04.2018, p. 278-287.

Research output: Contribution to journalArticle

@article{f521496bc3f649b2a2d7eda0124887a5,
title = "Bayesian Factor Model Shrinkage for Linear IV Regression With Many Instruments",
abstract = "A Bayesian approach for the many instruments problem in linear instrumental variable models is presented. The new approach has two components. First, a slice sampler is developed, which leverages a decomposition of the likelihood function that is a Bayesian analogue to two-stage least squares. The new sampler permits nonconjugate shrinkage priors to be implemented easily and efficiently. The new computational approach permits a Bayesian analysis of problems that were previously infeasible due to computational demands that scaled poorly in the number of regressors. Second, a new predictor-dependent shrinkage prior is developed specifically for the many instruments setting. The prior is constructed based on a factor model decomposition of the matrix of observed instruments, allowing many instruments to be incorporated into the analysis in a robust way. Features of the new method are illustrated via a simulation study and three empirical examples.",
keywords = "Bayesian econometrics, Horseshoe prior, Instrumental variables, Slice sampler",
author = "Paul Hahn and Jingyu He and Hedibert Lopes",
year = "2018",
month = "4",
day = "3",
doi = "10.1080/07350015.2016.1172968",
language = "English (US)",
volume = "36",
pages = "278--287",
journal = "Journal of Business and Economic Statistics",
issn = "0735-0015",
publisher = "American Statistical Association",
number = "2",

}

TY - JOUR

T1 - Bayesian Factor Model Shrinkage for Linear IV Regression With Many Instruments

AU - Hahn, Paul

AU - He, Jingyu

AU - Lopes, Hedibert

PY - 2018/4/3

Y1 - 2018/4/3

N2 - A Bayesian approach for the many instruments problem in linear instrumental variable models is presented. The new approach has two components. First, a slice sampler is developed, which leverages a decomposition of the likelihood function that is a Bayesian analogue to two-stage least squares. The new sampler permits nonconjugate shrinkage priors to be implemented easily and efficiently. The new computational approach permits a Bayesian analysis of problems that were previously infeasible due to computational demands that scaled poorly in the number of regressors. Second, a new predictor-dependent shrinkage prior is developed specifically for the many instruments setting. The prior is constructed based on a factor model decomposition of the matrix of observed instruments, allowing many instruments to be incorporated into the analysis in a robust way. Features of the new method are illustrated via a simulation study and three empirical examples.

AB - A Bayesian approach for the many instruments problem in linear instrumental variable models is presented. The new approach has two components. First, a slice sampler is developed, which leverages a decomposition of the likelihood function that is a Bayesian analogue to two-stage least squares. The new sampler permits nonconjugate shrinkage priors to be implemented easily and efficiently. The new computational approach permits a Bayesian analysis of problems that were previously infeasible due to computational demands that scaled poorly in the number of regressors. Second, a new predictor-dependent shrinkage prior is developed specifically for the many instruments setting. The prior is constructed based on a factor model decomposition of the matrix of observed instruments, allowing many instruments to be incorporated into the analysis in a robust way. Features of the new method are illustrated via a simulation study and three empirical examples.

KW - Bayesian econometrics

KW - Horseshoe prior

KW - Instrumental variables

KW - Slice sampler

UR - http://www.scopus.com/inward/record.url?scp=85018185509&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85018185509&partnerID=8YFLogxK

U2 - 10.1080/07350015.2016.1172968

DO - 10.1080/07350015.2016.1172968

M3 - Article

VL - 36

SP - 278

EP - 287

JO - Journal of Business and Economic Statistics

JF - Journal of Business and Economic Statistics

SN - 0735-0015

IS - 2

ER -