Location Properties of Point Estimators in Linear Instrumental Variables and Related Models

Keisuke Hirano, Jack R. Porter

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

We examine statistical models, including the workhorse linear instrumental variables model, in which the mapping from the reduced form distribution to the structural parameters of interest is singular. The singularity of this mapping implies certain fundamental restrictions on the finite sample properties of point estimators: they cannot be unbiased, quantile-unbiased, or translation equivariant. The nonexistence of unbiased estimators does not rule out bias reduction of standard estimators, but implies that the bias-variance tradeoff cannot be avoided and needs to be considered carefully. The results can also be extended to weak instrument asymptotics by using the limits of experiments framework.

Original languageEnglish (US)
Pages (from-to)720-733
Number of pages14
JournalEconometric Reviews
Volume34
Issue number6-10
DOIs
StatePublished - May 22 2015

Fingerprint

Instrumental variables
Estimator
Finite sample properties
Structural parameters
Reduced form
Experiment
Quantile
Weak instruments
Bias reduction
Trade-offs
Singularity
Statistical model

Keywords

  • Bias correction
  • Finite sample properties
  • Instrumental variables
  • Weak instruments

ASJC Scopus subject areas

  • Economics and Econometrics

Cite this

Location Properties of Point Estimators in Linear Instrumental Variables and Related Models. / Hirano, Keisuke; Porter, Jack R.

In: Econometric Reviews, Vol. 34, No. 6-10, 22.05.2015, p. 720-733.

Research output: Contribution to journalArticle

@article{89764b1e46e24de2936bc4ae2238195b,
title = "Location Properties of Point Estimators in Linear Instrumental Variables and Related Models",
abstract = "We examine statistical models, including the workhorse linear instrumental variables model, in which the mapping from the reduced form distribution to the structural parameters of interest is singular. The singularity of this mapping implies certain fundamental restrictions on the finite sample properties of point estimators: they cannot be unbiased, quantile-unbiased, or translation equivariant. The nonexistence of unbiased estimators does not rule out bias reduction of standard estimators, but implies that the bias-variance tradeoff cannot be avoided and needs to be considered carefully. The results can also be extended to weak instrument asymptotics by using the limits of experiments framework.",
keywords = "Bias correction, Finite sample properties, Instrumental variables, Weak instruments",
author = "Keisuke Hirano and Porter, {Jack R.}",
year = "2015",
month = "5",
day = "22",
doi = "10.1080/07474938.2014.956573",
language = "English (US)",
volume = "34",
pages = "720--733",
journal = "Econometric Reviews",
issn = "0747-4938",
publisher = "Taylor and Francis Ltd.",
number = "6-10",

}

TY - JOUR

T1 - Location Properties of Point Estimators in Linear Instrumental Variables and Related Models

AU - Hirano, Keisuke

AU - Porter, Jack R.

PY - 2015/5/22

Y1 - 2015/5/22

N2 - We examine statistical models, including the workhorse linear instrumental variables model, in which the mapping from the reduced form distribution to the structural parameters of interest is singular. The singularity of this mapping implies certain fundamental restrictions on the finite sample properties of point estimators: they cannot be unbiased, quantile-unbiased, or translation equivariant. The nonexistence of unbiased estimators does not rule out bias reduction of standard estimators, but implies that the bias-variance tradeoff cannot be avoided and needs to be considered carefully. The results can also be extended to weak instrument asymptotics by using the limits of experiments framework.

AB - We examine statistical models, including the workhorse linear instrumental variables model, in which the mapping from the reduced form distribution to the structural parameters of interest is singular. The singularity of this mapping implies certain fundamental restrictions on the finite sample properties of point estimators: they cannot be unbiased, quantile-unbiased, or translation equivariant. The nonexistence of unbiased estimators does not rule out bias reduction of standard estimators, but implies that the bias-variance tradeoff cannot be avoided and needs to be considered carefully. The results can also be extended to weak instrument asymptotics by using the limits of experiments framework.

KW - Bias correction

KW - Finite sample properties

KW - Instrumental variables

KW - Weak instruments

UR - http://www.scopus.com/inward/record.url?scp=84938505150&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84938505150&partnerID=8YFLogxK

U2 - 10.1080/07474938.2014.956573

DO - 10.1080/07474938.2014.956573

M3 - Article

AN - SCOPUS:84938505150

VL - 34

SP - 720

EP - 733

JO - Econometric Reviews

JF - Econometric Reviews

SN - 0747-4938

IS - 6-10

ER -