Jackknifing type weighted least squares estimators in partially linear regression models

摘要

In a heteroskedastic partially linear regression model, You and Chen (Technical Report, Department of Mathematics and Statistics, University of Regina, 2000) proposed a semiparametric generalized least squares estimator (SGLSE). In this paper, a jackknife-type estimator of the asymptotic covariance matrix of the SGLSE is proposed. It is shown that this jackknife-type estimator is consistent and performs better than the usual δ method in some cases.