Accounting for real exchange rate changes at long time horizons

Engel (1999) introduced real exchange rate accounting to determine the importance of nontradables for real exchange rate movements. We extend his approach in two directions. First, we identify a potential bias in the mean squared error (MSE) measure used in previous work. Second, using the corrected MSE measure we provide new empirical evidence that nontradables explain real exchange rate movements but only at really long horizons – over decades not years.