A comparison of different methods of estimating and testing the especification of rational expectation models with one endogeneus and one exogenus variable

This article considers the theory of the estimation and testing of a model with one endogenous variable and one exogenous variable, where the structure of the model assumes a simple rational expectations hypothesis for the determination of the endogenous variable. Two methods of estimation are considered, the first the method of Maximum Likelihood, and the second the method of Instrumental Variables. The first is asymptotically efficient, the second may be relatively less asymptotically efficient. The first also has the advantage of suggesting suitable tests for the general form of the rational expectations model