A Double Length Regression Computation Method for the 2SGLS Estimator of Rational Expectations Models.

This paper extends a matrix inverse result of M. L. Higgins (1994) and presents a new unified double length regression method to calculate the two-step generalized least squares estimators of two types of rational expectations models with current anticipated and unanticipated components. The estimator can be applied directly in most of the standard econometric computer packages such as PC-Give and Microfit. Copyright 1996 by Blackwell Publishing Ltd