This project propose to reconcile the results found in Andolfatto et al. (2008) andDel Negro and Eusepi (2011). The former shows that imperfect information model isable to incorporate important feature of the inflation expectation data, but it is notallowed by rational expectation models: correlated forecast errors. The latter, usingBayesian techniques, find that rational expectations models performs better than theimperfect information model in fitting the same inflation expectations data. They arguethat the imperfect information model imply a too restrictive learning process that is atthe odds with the data.We propose a model imperfect information model with heterogeneous information -Imperfect common knowledge model - that allows correlated forecast errors, but a lessrestricted learning process. We believe that this model can be a better framework tostudy inflation expectations. In order to solve such a model, we propose a new solutionmethod that incorporate state endogenous variables and public signals to the imperfectcommon knowledge models. The solution method may be useful for a wide range ofapplications using the imperfect common knowledge information structure.
News published in Agência FAPESP Newsletter about the scholarship: