Description

We introduce cognitive discounting into a full fledged monetary DSGE model to cope with the forward guidance puzzle and propose an estimation strategy that relies on system priors to guide the model into delivering data-consistent IRFs for monetary policy and forward guidance shocks. We find that the successful implementation of this behavioral hypothesis crucially hinges on allowing agents to entertain a cognitive discount factor specific for future monetary policy announcements. The estimated model attains a significantly better fit to the data than its rational expectations counterpart, while relying on only slightly modified estimates of structural parameters and substantial degrees of forward guidance discounting. Cognitive discounting of future events triggered by non-forward guidance shocks in our model is limited, and so is its contribution to the improvement of the marginal data density. We further find that professional forecasts are consistent with rational expectations and, thus, not appropriate for the estimation of forward guidance cognitive discounting.