Publisher:
Universität Dortmund, Sonderforschungsbereich 475 - Komplexitätsreduktion in Multivariaten Datenstrukturen, Dortmund
Abstract:
In this paper we consider the asymptotic distribution of S -estimators in the nonlinear regression model with long-memory error terms. S - estimators are robust estimates with a high breakdown point and good asymptotic properties in the i.i.d case. They are constructed for linear regression. In the nonlinear regression model with long-memory errors it turns out. that S-estimators are asymptotically normal with a rate of convergence of n1-h , ½ < H < 1. But the distribution depends heavily on the unknown parameter vector.