In this paper we consider the implications of a fully specified dynamic general equilibrium model, developed by Smets and Wouters (2003). This is a relatively large-scale forward-looking model, which was shown to provide a good fit to the data. We show that systematically accounting for prior uncertainty may lead to substantially different parameter estimates. However many of the qualitative features of the model remain similar under the alternative estimates that we find. We then formulate and analyze optimal policy rules in the model, focusing on a simple loss function which is commonly used and is independent of the estimates. We determine the optimal equilibrium dynamics for our estimates as well as those of Smets and Wouters, and find that they imply largely similar behavior. We then analyze simple policy rules, finding that these rules perform relatively well and are robust to our different sets of parameter estimates. Overall, our results suggest that the model may be relatively robust in its ability to capture certain aspects of the data. However some caution should be exercised in basing inference on the structural estimates, as these seem to be only weakly identified.
