On the Estimation of an Almost Ideal Demand System with Autoregressive Errors

Abstract

A new GLS estimation procedure is employed in the estimation of a dynamic version of Almost Ideal Demand System with vector autoregressive errors. The procedure enables us to have more natural form of autoregressive parameters. Also, a new habit formation specification obeying the adding up restriction is proposed. In an application for four categories of food in the U.S., test results show that the error process is a first order vector autoregressive process and that the demand system is characterized by habit formation.