A Note on Information Flows and Identification of News Shocks Models

This note points out a hitherto unrecognised identification issue in a class of rational expectations (RE) models with news shocks. We show that different degrees of anticipation (information flows) have strikingly different implications for the identifiability of the underlying structural model, irrespective of its non-fundamental time-series representation. In particular, under full shock anticipation equilibrium reduced forms behave as noisy perfect foresight state motions, which are non-identifiable. As a consequence, the underlying news shocks model fails to be (first-order) identified. The identification failure is illustrated with a New Keynesian model that can be solved analytically.