Rational Expectations and Learning
The rational expectations hypothesis assumes that agents fully utilize the information available to them and that they also know the parameters of the model that determines the variables about which they form expectations.
An alternative hypothesis is that agents learn over time about the values of the parameters of the model that determines the relevant variables. This learning approach to the modeling of expectations formation in macroeconomics assumes that agents adapt their expectations on the basis of new data.Learning models have some differences from rational expectations models. First, they provide a boundedly rational approach as to how full rational expectations can be reached. Second, learning acts as a selection device in models with multiple rational expectations equilibria. Third, the learning dynamics themselves may be of interest. Although there are various approaches to learning in macroeconomics, the emphasis usually is on adaptive learning schemes, in which agents use statistical or econometric techniques in self-referential stochastic systems.
This adaptive learning approach to the formation of expectations is surveyed by Evans and Honkapohja [1999] and is analyzed extensively in Evans and Honkapohja [2001].
9.7