A weighted pairwise likelihood approach to multivariate AR(1) models

Abstract

In this paper, the use of weighted pairwise likelihood instead of the full likelihood in estimating the parameters of the multivariate AR(1) is investigated. A closed formula for typical elements of the Godambe information(sandwich information) is presented. Some efficiency calculations are also given to discuss the feasibility andcomputational advantages of the weighted pairwise likelihood approach relative to the full likelihood approach.