Partial Linear Quantile Regression and Bootstrap Confidence Bands

Abstract

In this paper uniform confidence bands are constructed for nonparametric quantile estimates of regression functions. The method is based on the bootstrap, where resampling is done from a suitably estimated empirical density function (edf) for residuals. It is known that the approximation error for the uniform confidence band by the asymptotic Gumbel distribution is logarithmically slow. It is proved that the bootstrap approximation provides a substantial improvement. The case of multidimensional and discrete regressor variables is dealt with using a partial linear model. Comparison to classic asymptotic uniform bands is presented through a simulation study. An economic application considers the labour market discrimination effect with respect to different education levels. Keywords: Bootstrap, Quantile Regression, Confidence Bands, Nonparametric Fitting, Kernel Smoothing, Partial Linear Model