Economic Fundamentals and Short-Run Exchange Rate Prediction: A Machine-Learning Perspective

Ilias Filippou, David E. Rapach, Mark P. Taylor, and Guofu Zhou

♦ The exchange rate predictability puzzle—that fundamentals fail to forecast short-horizon exchange rate movements out of sample—has long resisted resolution. We show that fundamentals do matter, but their effects vary with global financial conditions. Using a rich set of country characteristics interacted with global state variables and machine-learning techniques to handle the high-dimensional data, we document that interactions between country characteristics and foreign exchange market volatility are particularly important, consistent with a “flight to fundamentals” during periods of financial stress. Our forecasts deliver substantial economic value and offer new insight into the carry trade.

Read it here.