Abstract:
We assess the stability of the unemployment gap parameter using linear dynamic Phillips curve models for the United States. In this study, we allow the unemployment gap parameter to be time-varying such that we can monitor the importance of the Phillips curve over time. We consider different specifications and different measures for inflation. Furthermore, we include stochastic volatility for the observation errors. Our estimation results are based on practical Bayesian state space methods which include feasible testing and diagnostic checking procedures. A key finding is that the Phillips curve for U.S. headline inflation has remained empirically relevant over the years.