Exchange rate forecasting: the errors weve really made

We examine the real-time forecasting performance of standard exchange rate models, using dozens of different vintages of data. Favorable evidence of long-horizon exchange rate predictability for the DM and Yen found in Mark (American Economic Review 1995;85:201–218) is present in only a two-year window of data vintages around that originally used. Approximately one-third of the improved forecasting performance over a random walk is eventually undone by data revisions. We also find the models consistently perform better using original release data than fully-revised data, and sometimes forecast better using realtime forecasts of future fundamentals instead of actual future fundamentals, contradicting a cherished presumption dating back to Meese and Rogoff (Journal of International Economics 1983;14:3–24).