Learning in Survey Expectations of Inflation
Krisztina Molnar (NHH, Bergen) (joint with Zoltan Reppa)
Abstract
A theoretical result in the theory of adaptive learning is that the optimal tracking parameter is larger if the underlying environment is more unstable. In other words when the underlying environment is subject to changes agents pay more attention to recent data and discount more past data when forming their expectations. In this paper we test this result using survey data on inflation expectations for 25 countries. We find that, in accordance with theory, the tracking parameter that best describes expectation formation is larger in countries that are more unstable. This result is robust to various measures of instability.