Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
This paper proposes a model where heterogeneous agents formulate their predictions of exchange rates based on a Bayesian learning process and higher-order beliefs where fundamentals and private information are used. We exploit survey data on professional forecasts to estimate the model through a Bayesian approach. Our analysis shows that higher-order beliefs are crucial, as they improve the ability to make predictions of exchange rates due to the possible coordination among agents. Moreover, public information plays the most critical role in determining individual predictions. Although the precision of the private signal is higher than the public one, information publicly revealed does exert a disproportionate influence, and differences in the estimated signals determine the equilibrium strategy of each agent as a combination of personal beliefs and higher-order expectations.