Generalized Least Squares Estimation of Linear Models Containing Rational Future Expectations.

The authors discuss the choice of approximations for unobserved expectations underlying consistent estimators in linear rational expectations models with future expectations. They show how estimators that are more efficient than the commonly used GMM estimators can be obtained if it is assumed that the future expectation depends on a finite number of variables only. Numerical results for a simple model illustrate the related efficiency of various estimators. Copyright 1991 by Economics Department of the University of Pennsylvania and the Osaka University Institute of Social and Economic Research Association.