Higher order beliefs and the dynamics of exchange rates

We estimate a model for exchange rate dynamics when expectations present higher order beliefs. A structural macro model for exchange rates is proposed where agents form their one-step-ahead predictions under a Bayesian learning process and in which aggregation of their choices is considered into the dynamics of exchange rates. Bayesian estimation of the structural parameters is implemented by using survey data on heterogeneous forecasts and fundamentals. Results show that higher order beliefs are relevant in building the subjective expectations' process and this phenomenon is dependent on information uncertainty. First, public information coordinates heterogeneous expectations of predictors leading to overweighting of signals from specific fundamentals. Second, predictors mis-perceive them as the most prevalent factors to form their forecasts, although the role of their own private assessments fades away.