摘要：he empirical literature has consistently rejected that the uncovered interest parity (UIP)
theorem holds in practice, thus posing the well-known forward premium puzzle. In this study,
we examine this issue for a sample of 18 emerging market currencies and, in addition, for a
subsample of 6 currencies from emerging Europe. We f irst conf irm earlier evidence for the
existence of a forward premium puzzle for emerging market economies. We then extend the
model with a view to exploring systematic relationships between excess returns from invest-
ments in foreign currency and country-specif ic economic fundamentals. Subsequently, we use
this extended model to generate out-of-sample forecasts of currency returns. We also test for
forecast accuracy, conf irming that these forecasts are superior to naïve forecasts. Our results
show that investments based on these forecasts generate considerably higher returns than
alternative investment strategies. This applies in particular to our full sample of 18 emerging
market currencies. For the subsample of 6 currencies from emerging Europe, prof its per trade
for the model-based forecasts also outperform those generated by the other investment strat-
egies, but by a smaller margin. These results suggest that, compared with currencies of
advanced countries, the smaller bias in the forward exchange rates of emerging market
currencies found in the empirical literature could relate to the better predictability of currency
returns for emerging market currencies.
