than the average misalignment in Poland, 3.6%, which, with a more flexible nominal exchange
rate, experienced no currency crisis at all.
9. Out of Sample Forecasting
Finally, can the movements of the macroeconomic fundamentals and the exogenous
variables predict the movements of the real exchange rate in the post crisis periods in Russia? To
answer this, two out of sample forecasts for the post-crisis period, based on the error correction
equation specified for the pre-crisis periods (specification 4 of Table 8), are performed. First, the
real exchange rates are forecast for the post crisis periods with the known values of the
exogenous variables and lagged values of the real exchange rate -- a one period ahead static
forecast. And second, the forecasts are performed with the lagged values of the real exchange
rate from the forecast of the real exchange rate in the previous period and with the known values
of the exogenous variables -- a one period ahead dynamic forecast.
The results are presented in Figure 5 and Table 10. The standard measures of forecast errors
indicate that the static forecasts do reasonably well. This suggests that the error correction
equation specified for the pre-crisis period may be generalizable. However, the dynamic
forecasts yield much less satisfactory results. The predicted path entirely misses the movements
of the real exchange rate, indicating that the values of the macroeconomic fundamentals in the
pre-crisis period are not able to explain the movements of the real exchange rate in the post-crisis
period, demonstrating the severity of the crisis and the strength of speculative factors not
included in the model. The misalignment calculated on the basis of macro fundamentals, quite
severe prior to August 1998 (Figure 4), may be seen as a precondition for the crisis, but we are
not able to predict the crisis per se.
40