ESTIMATION, INFERENCE, AND SPECIFICATION TESTING FOR POSSIBLY MISSPECIFIED QUANTILE REGRESSION

Tae Hwan Kim; Halbert White

doi:10.1016/S0731-9053(03)17005-3

ESTIMATION, INFERENCE, AND SPECIFICATION TESTING FOR POSSIBLY MISSPECIFIED QUANTILE REGRESSION

Tae Hwan Kim, Halbert White

Department of Economics

Research output: Chapter in Book/Report/Conference proceeding › Chapter

75 Citations (Scopus)

Abstract

To date, the literature on quantile regression and least absolute deviation regression has assumed either explicitly or implicitly that the conditional quantile regression model is correctly specified. When the model is misspecified, confidence intervals and hypothesis tests based on the conventional covariance matrix are invalid. Although misspecification is a generic phenomenon and correct specification is rare in reality, there has to date been no theory proposed for inference when a conditional quantile model may be misspecified. In this paper, we allow for possible misspecification of a linear conditional quantile regression model. We obtain consistency of the quantile estimator for certain "pseudo-true" parameter values and asymptotic normality of the quantile estimator when the model is misspecified. In this case, the asymptotic covariance matrix has a novel form, not seen in earlier work, and we provide a consistent estimator of the asymptotic covariance matrix. We also propose a quick and simple test for conditional quantile misspecification based on the quantile residuals.

Original language	English
Title of host publication	Maximum Likelihood Estimation of Misspecified Models
Subtitle of host publication	Twenty Years Later
Publisher	JAI Press
Pages	107-132
Number of pages	26
ISBN (Print)	0762310758, 9780762310753
DOIs	https://doi.org/10.1016/S0731-9053(03)17005-3
Publication status	Published - 2003

Publication series

Name	Advances in Econometrics
Volume	17
ISSN (Print)	0731-9053

Bibliographical note

Funding Information:
The authors would like to thank Douglas Stone for providing the data used in the paper and Clive Granger, Patrick Fitzsimmons, Alex Kane, Paul Newbold, Christophe Muller, Christian Gourieroux, Adrian Pagan, and Jin Seo Cho for their helpful comments. White’s research was supported by NSF grants SBR-9811562 and SES-0111238 and Kim’s research was supported by a grant from the University of Nottingham.

All Science Journal Classification (ASJC) codes

Economics and Econometrics

Access to Document

10.1016/S0731-9053(03)17005-3

Cite this

@inbook{0adb70505c8b430c9901c179bc3c4046,

title = "ESTIMATION, INFERENCE, AND SPECIFICATION TESTING FOR POSSIBLY MISSPECIFIED QUANTILE REGRESSION",

abstract = "To date, the literature on quantile regression and least absolute deviation regression has assumed either explicitly or implicitly that the conditional quantile regression model is correctly specified. When the model is misspecified, confidence intervals and hypothesis tests based on the conventional covariance matrix are invalid. Although misspecification is a generic phenomenon and correct specification is rare in reality, there has to date been no theory proposed for inference when a conditional quantile model may be misspecified. In this paper, we allow for possible misspecification of a linear conditional quantile regression model. We obtain consistency of the quantile estimator for certain {"}pseudo-true{"} parameter values and asymptotic normality of the quantile estimator when the model is misspecified. In this case, the asymptotic covariance matrix has a novel form, not seen in earlier work, and we provide a consistent estimator of the asymptotic covariance matrix. We also propose a quick and simple test for conditional quantile misspecification based on the quantile residuals.",

author = "Kim, {Tae Hwan} and Halbert White",

note = "Funding Information: The authors would like to thank Douglas Stone for providing the data used in the paper and Clive Granger, Patrick Fitzsimmons, Alex Kane, Paul Newbold, Christophe Muller, Christian Gourieroux, Adrian Pagan, and Jin Seo Cho for their helpful comments. White{\textquoteright}s research was supported by NSF grants SBR-9811562 and SES-0111238 and Kim{\textquoteright}s research was supported by a grant from the University of Nottingham. ",

year = "2003",

doi = "10.1016/S0731-9053(03)17005-3",

language = "English",

isbn = "0762310758",

series = "Advances in Econometrics",

publisher = "JAI Press",

pages = "107--132",

booktitle = "Maximum Likelihood Estimation of Misspecified Models",

address = "United States",

}

TY - CHAP

T1 - ESTIMATION, INFERENCE, AND SPECIFICATION TESTING FOR POSSIBLY MISSPECIFIED QUANTILE REGRESSION

AU - Kim, Tae Hwan

AU - White, Halbert

N1 - Funding Information: The authors would like to thank Douglas Stone for providing the data used in the paper and Clive Granger, Patrick Fitzsimmons, Alex Kane, Paul Newbold, Christophe Muller, Christian Gourieroux, Adrian Pagan, and Jin Seo Cho for their helpful comments. White’s research was supported by NSF grants SBR-9811562 and SES-0111238 and Kim’s research was supported by a grant from the University of Nottingham.

PY - 2003

Y1 - 2003

N2 - To date, the literature on quantile regression and least absolute deviation regression has assumed either explicitly or implicitly that the conditional quantile regression model is correctly specified. When the model is misspecified, confidence intervals and hypothesis tests based on the conventional covariance matrix are invalid. Although misspecification is a generic phenomenon and correct specification is rare in reality, there has to date been no theory proposed for inference when a conditional quantile model may be misspecified. In this paper, we allow for possible misspecification of a linear conditional quantile regression model. We obtain consistency of the quantile estimator for certain "pseudo-true" parameter values and asymptotic normality of the quantile estimator when the model is misspecified. In this case, the asymptotic covariance matrix has a novel form, not seen in earlier work, and we provide a consistent estimator of the asymptotic covariance matrix. We also propose a quick and simple test for conditional quantile misspecification based on the quantile residuals.

AB - To date, the literature on quantile regression and least absolute deviation regression has assumed either explicitly or implicitly that the conditional quantile regression model is correctly specified. When the model is misspecified, confidence intervals and hypothesis tests based on the conventional covariance matrix are invalid. Although misspecification is a generic phenomenon and correct specification is rare in reality, there has to date been no theory proposed for inference when a conditional quantile model may be misspecified. In this paper, we allow for possible misspecification of a linear conditional quantile regression model. We obtain consistency of the quantile estimator for certain "pseudo-true" parameter values and asymptotic normality of the quantile estimator when the model is misspecified. In this case, the asymptotic covariance matrix has a novel form, not seen in earlier work, and we provide a consistent estimator of the asymptotic covariance matrix. We also propose a quick and simple test for conditional quantile misspecification based on the quantile residuals.

UR - http://www.scopus.com/inward/record.url?scp=33645021852&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33645021852&partnerID=8YFLogxK

U2 - 10.1016/S0731-9053(03)17005-3

DO - 10.1016/S0731-9053(03)17005-3

M3 - Chapter

AN - SCOPUS:33645021852

SN - 0762310758

SN - 9780762310753

T3 - Advances in Econometrics

SP - 107

EP - 132

BT - Maximum Likelihood Estimation of Misspecified Models

PB - JAI Press

ER -

ESTIMATION, INFERENCE, AND SPECIFICATION TESTING FOR POSSIBLY MISSPECIFIED QUANTILE REGRESSION

Abstract

Publication series

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this