CAN LONG HORIZON DATA BEAT RANDOM WALK UNDER ENGEL-WEST EXPLANATION?

Engel and West (2004a) provide an explanation to reconcile the random walk behavior of exchange rate and linear present value asset pricing models. In this paper, we study the long horizon property of exchange rate under Engel-West explanation. It is found that the long horizon data can not significantly improve our chance of beating random walk. This result is consistent with recent empirical studies on the long horizon exchange rate. Under E-W explanation, the change of exchange rate can be more serially correlated in the long horizon data, but this change in most cases is only marginal. Depending on the persistence of change in fundamentals, two patterns may exist between the autocorrelation of exchange rate change and the time horizon. Both of these two patterns are found existing in the real data of exchange rates. These results support E-W explanation for exchange rate puzzle.