Implications of Asset Market Data for Equilibrium Models of Exchange Rates

When investors can trade home and foreign currency risk-free bonds, the exchange rate conditionally appreciates in states of the world that are worse for home investors than foreign investors. This prediction is at odds with the empirical evidence. We show that it can be overturned unconditionally if the deviations from U.I.P. are large and exchange rates are highly predictable, which are conditions that do not hold in the data. More broadly, it is not possible to match the empirical exchange rate cyclicality (the Backus-Smith puzzle), the deviations from U.I.P. (the Fama puzzle) and the lack of predictability (the Meese-Rogoff puzzle). Our minimalist approach allows us to delineate the class of models that can resolve these puzzles. We show that introducing Euler equation wedges consistent with a home currency bias, home bond convenience yields/financial repression, or intermediary frictions can resolve all of these puzzles.