Sequential action and beliefs under partially observable DSGE environments

Abstract

This paper introduces a classification of DSGEs from a Markovian perspective, and positions the class of POMDP (Partially Observable Markov Decision Process) to the center of a generalization of linear rational expectations models. The analysis of the POMDP class builds on the previous development in dynamic controls for linear system, and derives a solution algorithm by formulating an equilibrium as a fixed point of an operator that maps what we observe into what we believe.