Nonparametric Regression with Spatially Dependent Data
39 Pages Posted: 2 Oct 2009 Last revised: 1 May 2012
Date Written: March 30, 2010
In this paper we present a new procedure for nonparametric regression in case of spatially dependent data. In particular, we extend usual local linear regression (along the lines of Martins-Filho and Yao, 2009) and propose a two-step method where information on spatial dependence is incorporated in the error covariance matrix, estimated nonparametrically. The finite sample performance of our proposed procedure is then shown via Monte Carlo simulations for various data generating processes.
Keywords: nonparametric smoothing, spatial dependence
JEL Classification: C14, C21
Suggested Citation: Suggested Citation