Bootstrap confidence bands and partial linear quantile regression

Abstract

In this paper bootstrap confidence bands are constructed for nonparametric quantile estimates of regression functions, where resampling is done from a suitably estimated empirical distribution function (edf) for residuals. It is known that the approximation error for the confidence band by the asymptotic Gumbel distribution is logarithmically slow. It is proved that the bootstrap approximation provides an improvement. The case of multidimensional and discrete regressor variables is dealt with using a partial linear model. An economic application considers the labor market differential effect with respect to different education levels.