The Asymptotic Behavior of Monotone Percentile Regression Estimates

摘要

The least absolute deviations estimate of a monotone regression function was derived by Robertson and Waltman (1968). Assuming the observation points become dense in the domain of the regression function, Casady and Cryer (1976) obtained an upper bound on the almost sure rate of convergence of the estimator. The asymptotic distribution of the estimator at a point is obtained here and a faster rate of convergence is obtained. (Author)