This paper builds a two-country dynamic stochastic general equilibrium macro model to understand three empirical facts about international currency returns. They are the downward forward premium bias, the carry trade return, and the long-run risk reversal. Cross-country heterogeneity in unit-root productivity levels generates the systematic risk priced into currency returns. The risk can be magnified through monetary policy. Both a complete markets and an incomplete markets model are qualitatively consistent with these facts. Quantitatively, the incomplete markets model performs better.