Does the forward premium anomaly depend on the sample period used or on the sign of the premium?

Some recent studies claim that forward exchange rate bias is asymmetrical in that the empirical rejection of the unbiasedness hypothesis, especially the forward premium anomaly (i.e., the finding of a significantly negative relation between the forward premium and the future change in the spot rate), is for the situation when the forward U.S. dollar is quoted at a discount but not for the cases when it is quoted at a premium. At the same time, some other studies show that the forward premium anomaly is mainly associated with the data of the early and mid-1980s. This study demonstrates that the forward premium anomaly mainly depends on the sample period used, rather than on the sign of the forward premium. Evidence provided in this paper suggests either a significantly negative relation between the forward premium and the future change in the spot rate in the 1980s, or simply no significant relation between the two during other periods regardless of whether the forward dollar is quoted at a premium or a discount.