Description

The idea of “anchored” inflation expectations is often understood as a situation in which long-run expected inflation does not significantly respond to new information. Furthermore, expectations are thought to become “unanchored” only after a long enough sequence of inflation surprises. In this paper we conceptualize this idea in a monetary DSGE model with a time-varying learning mechanism, in which the sensitivity of agents to incoming data depends on accumulated inflation forecast errors. The latter affect the learning gain that agents use to update their beliefs on future inflation. We show how this mechanism improves the fit of the model to macroeconomic data, including expected inflation, for the Chilean inflation targeting period. In particular, we show that observed episodes with anchored and unanchored expectations are well captured by the estimated time-varying learning gain. We then use the estimated model to assess the role of monetary policy to anchor inflation expectations over time.