Generalization of the Kalman Filter for a kind of rational expectations models

Abstract

In this papel we obtain a generalization of the Kalman Filter for a kind of models in which the value of the vector variable in period t is explained linearly by the value it had in the previous period, by the pational expectations about the value that the variable y would take in period t, that the economic agents had in previous periods and by additive Gaussian noise. Then we try to get rid of the Gaussian hypothesis and we find a kind of systems in which we don't need that hypothesis, although these systems will not be, in general, rational expectations models.