Some answers to puzzles in testing unbiasedness in the foreign exchange market

Similar but alternative specifications of tests of forward rate unbiasedness provide conflicting evidence on the rejection of the hypothesis. These conflicting results are reconciled by demonstrating that although the root cause is simultaneity bias, the severity of this bias and the resulting rejection or non-rejection of the hypothesis depends entirely on the relative error variance empirical regularity common to foreign exchange markets. The analysis presented here applies to both stationary and non-stationary specifications of the model.