Inflation expectations and stability in an overlapping generations experiment with money creation

Abstract

We investigate how non-specialists form inflation expectations by running an experiment using a basic Overlapping Generations (OLG) model. The participants of the experiment are students of the University of Amsterdam, who predict inflation during 50 successive periods and are rewarded based on their accuracy. We include a central bank in the OLG model which increases the money supply at a constant rate. Participants are placed in separate OLG economies and are divided over two treatments: one with a "low" and one with a "high" money supply growth. We find that participants in the second treatment have substantially more difficulty in stabilizing inflation development by submitting accurate predictions than participants in the first treatment. However, when linear prediction rules are estimated on individual predictions, there is little difference between the two treatments. In both treatments, the most popular rules are Fundamentalist Expectations (predictions equal to the inflation sample mean) and Focal Expectations (predictions equal to a constant close to equilibrium). To verify whether participants adjust their prediction rules during the experiment, the estimated rules are checked for structural breaks. We find a surprisingly small number of structural breaks in both treatments.