DSGE Model Validation in a Bayesian Framework: an Assessment

This paper presents the concept of Model Validation applied to a Dynamic Stochastic General equilibrium Model (DSGE). The main problem discussed is the approximation of the statistical representation for a DSGE model when not all endogenous variables are observable. MonteCarlo experiments in artificial world are implemented to assess this problem by using the DSGE-VAR. Two Data Generating Processes are compared: a forward-looking and a backward-looking model. These experiments are followed by an empirical analysis with real world data for the US economy.