Heterogeneous Expectations and Uncertain Inflation Target

We analyze a new Keynesian economy populated by adaptive-learning agents with heterogeneous beliefs about the time-varying inflation target. A fraction of agents is assumed to have a full and updated information set including the permanent and tem- porary component of the inflation target at the current period, while the remainder of agents receives a signal and use it to estimate the target components solving a Kalman filter problem. The proportion of the two strategies is endogenous and depends on a measure of past performance of predictors. We conduct stochastic simulations to assess whether different hypotheses about the information regime may affect macroe- conomic stability in the short and in the long run. We find that a smaller proportion of agents using costly information is associated to larger expected losses. Nevertheless, the fraction of agents following this strategy drops signficantly in the aftermath of a shock to the inflation target because the Kalman signal extraction procedure allows to follow more closely the actual dynamics of the economy.