Bayesian forecasting of real exchange rates with a Dornbusch prior

This paper assesses if a Bayesian VAR with a Dornbusch prior outperforms the random walk model in predicting real exchange rates. Our main contributions are twofold. First, from a methodological point of view we apply an innovative framework to estimate structural Bayesian VAR models. Second, we provide evidence that a VAR with a Dornbusch prior can generate more accurate forecasts for real exchange rates than a standard VAR model based on the random walk prior and the naïve random walk model.