Multivariate Local Polynomial Regression With Autocorrelated Errors

We propose a three-step local polynomial procedure for a multivariate nonparametric regression in which the errors are autocorrelated. The proposed estimator uses all sample points to estimate m(x), the regression function evaluated at point x, but the contributions from all non-local points are used only through their residuals. Our proposed estimator exhibits good finite sample performance in a Monte Carlo simulation study.

MoreLess

Year of publication:	2012
Authors:	Yang, Ke
Published in:	Economics Bulletin. - AccessEcon, ISSN 1545-2921. - Vol. 32.2012, 4, p. 3298-3305
Publisher:	AccessEcon
Subject:	local polynomial regression \| autocorrelated errors \| efficiency

More details

Extent:	application/pdf
Type of publication:	Article
Classification:	C1 - Econometric and Statistical Methods: General ; C8 - Data Collection and Data Estimation Methodology; Computer Programs
Source:	RePEc - Research Papers in Economics

Persistent link: https://ebvufind01.dmz1.zbw.eu/10011278528