Abstract: We propose a varying-coefficient quantile regression model for survival data subject to random censoring. Motivated by the work of Yang (1999), quantile-based moments are constructed using covariate-weighted empirical cumulative hazard functions. We estimate regression parameters based on the generalized method of moments. The proposed estimators are shown to be consistent and asymptotically normal. We examine the proposed method with finite sample sizes through simulation studies, and illustrate it with a Richter’s syndrome study.

Key words and phrases: Generalized method of moments, local polynomial, regression quantiles, semiparametric models, random censoring, survival data.