Local Linear Estimation for Regression Models with Locally Stationary Long Memory Errors

Wang Lihong
DOI: https://doi.org/10.1016/j.jkss.2015.12.005
2016-01-01
Abstract:In this paper we consider the local linear regression estimation for the nonparametric regression models with locally stationary long memory errors. The asymptotic behaviors of the regression estimators are established. It is shown that there is a multiple bandwidth dichotomy for the asymptotic distribution of the estimators of the regression function and its derivatives. The finite sample performance of the estimator is discussed through simulation studies.
What problem does this paper attempt to address?