Bounding endogenous regressor coefficients using moment inequalities and generalized instruments

Jin Young Choi, Myoung-jae Lee

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

The main approach to deal with regressor endogeneity is instrumental variable estimator (IVE), where an instrumental variable (IV) m is required to be uncorrelated to the regression model error term u (COR(m,u)=0) and correlated to the endogenous regressor. If COR(m,u)≠0 is likely, then m gets discarded. But even when COR(m,u)≠0, often one has a good idea on the sign of COR(m,u). This article shows how to make use of the sign information on COR(m,u) to obtain an one-sided bound on the endogenous regressor coefficient, calling m a 'generalized instrument' or 'generalized instrumental variable (GIV)'. If there are two GIV's m 1 and m 2, then a two-sided bound or an improved one-sided bound can be obtained. Our approach is simple, needing only IVE; no non-parametrics, nor any 'tuning constants'. Specifically, the usual IVE is carried out, and the only necessary modification is that the estimate for the endogenous regressor coefficient is interpreted as a lower/upper bound depending on the prior notion on the sign of COR(m,u) and some estimable moment. A real data application is done to Korean household data with two or more children to illustrate our approach for the issue of child quantity-quality trade-off.

Original languageEnglish
Pages (from-to)161-182
Number of pages22
JournalStatistica Neerlandica
Volume66
Issue number2
DOIs
Publication statusPublished - 2012 May 1

Fingerprint

Coefficient Inequalities
Moment Inequalities
Instrumental Variables
Estimator
Endogeneity
Model Error
Coefficient
Error term
Tuning
Regression Model
Trade-offs
Likely
Instrumental variables estimator
Moment inequalities
Coefficients
Endogenous regressors
Upper bound
Moment
Necessary
Instrumental variables

Keywords

  • Bounding parameters
  • Generalized instrument
  • Moment inequalities

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Bounding endogenous regressor coefficients using moment inequalities and generalized instruments. / Choi, Jin Young; Lee, Myoung-jae.

In: Statistica Neerlandica, Vol. 66, No. 2, 01.05.2012, p. 161-182.

Research output: Contribution to journalArticle

@article{a8ac191398c044e3a8588d516bc3e0be,
title = "Bounding endogenous regressor coefficients using moment inequalities and generalized instruments",
abstract = "The main approach to deal with regressor endogeneity is instrumental variable estimator (IVE), where an instrumental variable (IV) m is required to be uncorrelated to the regression model error term u (COR(m,u)=0) and correlated to the endogenous regressor. If COR(m,u)≠0 is likely, then m gets discarded. But even when COR(m,u)≠0, often one has a good idea on the sign of COR(m,u). This article shows how to make use of the sign information on COR(m,u) to obtain an one-sided bound on the endogenous regressor coefficient, calling m a 'generalized instrument' or 'generalized instrumental variable (GIV)'. If there are two GIV's m 1 and m 2, then a two-sided bound or an improved one-sided bound can be obtained. Our approach is simple, needing only IVE; no non-parametrics, nor any 'tuning constants'. Specifically, the usual IVE is carried out, and the only necessary modification is that the estimate for the endogenous regressor coefficient is interpreted as a lower/upper bound depending on the prior notion on the sign of COR(m,u) and some estimable moment. A real data application is done to Korean household data with two or more children to illustrate our approach for the issue of child quantity-quality trade-off.",
keywords = "Bounding parameters, Generalized instrument, Moment inequalities",
author = "Choi, {Jin Young} and Myoung-jae Lee",
year = "2012",
month = "5",
day = "1",
doi = "10.1111/j.1467-9574.2011.00501.x",
language = "English",
volume = "66",
pages = "161--182",
journal = "Statistica Neerlandica",
issn = "0039-0402",
publisher = "Wiley-Blackwell",
number = "2",

}

TY - JOUR

T1 - Bounding endogenous regressor coefficients using moment inequalities and generalized instruments

AU - Choi, Jin Young

AU - Lee, Myoung-jae

PY - 2012/5/1

Y1 - 2012/5/1

N2 - The main approach to deal with regressor endogeneity is instrumental variable estimator (IVE), where an instrumental variable (IV) m is required to be uncorrelated to the regression model error term u (COR(m,u)=0) and correlated to the endogenous regressor. If COR(m,u)≠0 is likely, then m gets discarded. But even when COR(m,u)≠0, often one has a good idea on the sign of COR(m,u). This article shows how to make use of the sign information on COR(m,u) to obtain an one-sided bound on the endogenous regressor coefficient, calling m a 'generalized instrument' or 'generalized instrumental variable (GIV)'. If there are two GIV's m 1 and m 2, then a two-sided bound or an improved one-sided bound can be obtained. Our approach is simple, needing only IVE; no non-parametrics, nor any 'tuning constants'. Specifically, the usual IVE is carried out, and the only necessary modification is that the estimate for the endogenous regressor coefficient is interpreted as a lower/upper bound depending on the prior notion on the sign of COR(m,u) and some estimable moment. A real data application is done to Korean household data with two or more children to illustrate our approach for the issue of child quantity-quality trade-off.

AB - The main approach to deal with regressor endogeneity is instrumental variable estimator (IVE), where an instrumental variable (IV) m is required to be uncorrelated to the regression model error term u (COR(m,u)=0) and correlated to the endogenous regressor. If COR(m,u)≠0 is likely, then m gets discarded. But even when COR(m,u)≠0, often one has a good idea on the sign of COR(m,u). This article shows how to make use of the sign information on COR(m,u) to obtain an one-sided bound on the endogenous regressor coefficient, calling m a 'generalized instrument' or 'generalized instrumental variable (GIV)'. If there are two GIV's m 1 and m 2, then a two-sided bound or an improved one-sided bound can be obtained. Our approach is simple, needing only IVE; no non-parametrics, nor any 'tuning constants'. Specifically, the usual IVE is carried out, and the only necessary modification is that the estimate for the endogenous regressor coefficient is interpreted as a lower/upper bound depending on the prior notion on the sign of COR(m,u) and some estimable moment. A real data application is done to Korean household data with two or more children to illustrate our approach for the issue of child quantity-quality trade-off.

KW - Bounding parameters

KW - Generalized instrument

KW - Moment inequalities

UR - http://www.scopus.com/inward/record.url?scp=84859758580&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84859758580&partnerID=8YFLogxK

U2 - 10.1111/j.1467-9574.2011.00501.x

DO - 10.1111/j.1467-9574.2011.00501.x

M3 - Article

AN - SCOPUS:84859758580

VL - 66

SP - 161

EP - 182

JO - Statistica Neerlandica

JF - Statistica Neerlandica

SN - 0039-0402

IS - 2

ER -