Smoothed Empirical Likelihood Methods for Quantile Regression Models

The standard confidence regions based on the first-order approximation of quantile regression estimators can be inaccurate in small samples. We show that confidence regions based on the smoothed empirical likelihood ratio have coverage errors of order n^{-1} and may be Bartlett-corrected to produce regions with an error of order n^{-2}, where n denotes the sample size. We further extend these results to censored quantile regression models. Our results are extensions of the previous results of Chen and Hall (1993) to the regression contexts. Also, from the duality of confidence regions and hypothesis tess, our results imply that the smoothed empirical likelihood confidence regions might be more accurate in small samples than the confidence regions that can be constructed from the smoothed bootstrap method recently suggested by Horowitz (1998).