Predictive Power of U.S. Macroeconomic Factors for the Dollar/Won Real Exchange Rate

This paper examines the asymmetric out-of-sample predictability of macroeconomic variables for the real exchange rate between the United States and Korea. While conventional models suggest that the bilateral real exchange rate is driven by the relative economic performance of the two countries, our research demonstrates the superior predictive power of our factor-augmented forecasting models only when factors are obtained from U.S. economic variables, whereas the inclusion of Korean factors fails to enhance predictability and behaves more like noise variables. We attribute the remarkable predictability of American factors to the significant cross-correlations observed among bilateral real exchange rates vis-à -vis the U.S. dollar, which suggests a limited influence of idiosyncratic factors specific to small countries. Moreover, we assess our factor-augmented forecasting models by incorporating proposition-based factors instead of macro factors. While macro factors generally exhibit superior performance, it is worth noting that the uncovered interest parity (UIP)-based global factors, with the dollar as the numéraire, consistently demonstrate strong overall performance. On the other hand, the purchasing power parity (PPP) and real uncovered interest parity (RIRP) factors have a limited role in forecasting the dollar/won real exchange rate. Our major findings are grounded in pre-COVID-19 era data, highlighting key insights drawn from periods of relative tranquility. We explore how economic crises act as catalysts, precipitating a separation of the real exchange rate from macroeconomic fundamentals.