A re-examination of the unbiasedness forward rate hypothesis using dynamic SUR model

This paper augments the empirical literature by testing the unbiasedness hypothesis of the U.S. foreign exchange market in a panel-based framework. Specifically, the dynamic seemingly unrelated estimator (Mark, Ogaki, & Sul, 2000) is applied to test parameter restrictions and cointegration, in additional to accounting for the cross-currency dependence. Evidence from daily data of 17 currencies indicates that although cointegration is confirmed, both the tests of parameter restrictions and the presence of serial correlation do not favor the unbiasedness hypothesis satisfactorily.