Read the full-text of this article for free

Title: Estimation After Pre-Testing in Least Absolute Value Regression with Autocorrelated Errors

Authors: Terry E. Dielman; Elizabeth L. Rose

Addresses: Author address listing can be found in the "About the Authors" section at the end of the article.

Abstract: We study least absolute value (LAV) estimation and inference in the context of simple time series regression when the disturbances are autocorrelated. Several different estimation techniques are compared: uncorrected LAV; LAV after a Cochrane-Orcutt (CO)-type transformation to correct for autocorrelation; LAV after a Prais-Winsten (PW)-type transformation to correct for autocorrelation; and two pre-test estimators that transform (by CO and by PW, respectively) when a pre-test suggests that autocorrelation is present. Monte Carlo simulation methods are used to compare the small-sample performances of the different estimators. The Prais-Winsten approach to correction for autocorrelation is preferable to the Cochrane-Orcutt approach, and there appears to be minimal loss associated with always correcting.

Keywords: Least absolute value (LAV); autocorrelated errors; autocorrelation; Monte Carlo simulation; Prais-Winsten approach; Cochrane-Orcutt approach.

DOI: 10.1504/JBM.1995.141004

Journal of Business and Management, 1995 Vol.2 No.2, pp.74 - 95

Published online: 05 Sep 2024 *