Extreme support for uncovered interest parity

Concerning UIP, the common conclusion is that it may be valid but undetectable for many reasons. In this paper we take a complementary route in that we base our methodology on a random time effects panel model that controls for various biasing factors and which is invariant to the choice of the numeraire currency. We show that the rejection of UIP is not as severe as is commonly found and that it almost perfectly holds in periods where the forward premiums are large.