A well-known puzzle in international finance is that a random walk predicts exchange rates better than economic models. I offer a potential explanation. When exchange rates and fundamentals are highly persistent, long-horizon forecasts of economic models are biased by the estimation error. When...