Fixed design regression quantiles for time series

This paper studies nonparametric estimation of regression quantiles under the fixed design model. We suppose that the error random variables are coming from a strictly stationary stochastic process satisfying the strong mixing condition. The joint asymptotic normality for the estimators of several quantiles is given. The same property is established for the regression quantile estimator at different fixed design points. (C) 2003 Elsevier B.V. All rights reserved.