Resolving the unbiasedness puzzle in the foreign exchange market

An unresolved puzzle in the empirical foreign exchange literature is that tests of forward rate unbiasedness using the forward rate and forward premium equations yield markedly different conclusions about the unbiasedness of the forward exchange rate. This puzzle is resolved by showing that because of the persistence in exchange rates, estimates of the slope coefficient from the forward premium equation are extremely sensitive to small violations of the null hypothesis of the type and magnitude that are likely to exist in the real world. Moreover, contrary to suggestions in the literature and common practice, the forward premium equation does not necessarily provide a better test of unbiasedness than the forward rate equation.