Bias-Corrected Quantile Regression Estimation of Censored Regression Models

Research output: Working paperDiscussion paperOther research output

559 Downloads (Pure)

Abstract

Motivated by weak small-sample performance of the censored regression quantile estimator proposed by Powell (1986a), two- and three-step estimation methods were introduced for estimation of the censored regression model under conditional quantile restriction. While those stepwise estimators have been proven to be consistent and asymptotically normal, their finite sample performance greatly depends on the specification of an initial estimator that selects the subsample to be used in subsequent steps. In this paper, an alternative semiparametric estimator is introduced that does not involve a selection procedure in the first step. The proposed estimator is based on the indirect inference principle and is shown to be consistent and asymptotically normal under appropriate regularity conditions. Its performance is demonstrated and compared to existing methods by means of Monte Carlo simulations.
Original languageEnglish
Place of PublicationTilburg
PublisherEconometrics
Number of pages45
Volume2014-060
Publication statusPublished - 6 Oct 2014

Publication series

NameCentER Discussion Paper
Volume2014-060

    Fingerprint

Keywords

  • asymptotic normality
  • censored regression
  • indirect inference
  • quantile regression

Cite this

Cizek, P., & Sadikoglu, S. (2014). Bias-Corrected Quantile Regression Estimation of Censored Regression Models. (CentER Discussion Paper; Vol. 2014-060). Tilburg: Econometrics.