Learning, Heuristics and Anchored Inflation: How Do Different Types of Consumer Change Their Minds about Inflation?

Alternative models of actual and expected inflation are estimated to investigate how, in real time and as information arrives, different types of consumer learn about inflation. The models distinguish various views about learning including those based on Bayesian updating or on learning-rigidity heuristics. Using a detailed Australian survey, we find that learning is best characterised by periodic shifts in beliefs rather than incremental Bayesian adjustments, that different consumer groups learn at different times and from different experiences and that the inflation expectations of different groups are anchored on a range of reference inflation rates which do not move in sync.