The Term Structure of Forward Exchange Rates and the Forecastability of Spot Exchange Rates: Correcting the Errors

This paper revisits one of the oldest questions in international finance: does the forward exchange rate contain useful information about of the future path of the spot exchange rate? We present a theoretical framework and provide evidence that challenges the common view (Mussa (1979); Dornbusch (1980); Frenkel (1981); Cumby-Obstfeld (1984) that forward premia contain little information regarding subsequent changes in the spot exchange rate. Using weekly dollar-DM and dollar-sterling data on spot exchange rates and 1, 3, 6., and 12 month forward exchange rates, we find that, as predicted by the theoretical framework the term structure of forward exchange rates together with the spot exchange rate comprise a system that is well represented by a vector vector correction model. Employing Johansen's (1991) maximum likelihood approach, we test and confirm for each country the existence of 4 cointegrating relationships as predicted by the theory. We then test and confirm for each country the joint hypothesis that a basis for this cointegrating space is the vector of 4 forward premia. We next test, and reject for each country, the hypothesis that the spot exchange rate is weekly exogenous with respect to the term structure of forward rates. Out-of-sample simulations indicate that the information contained in the term structure of forward premia can be used to reduce the mean squared error in forecasting the spot rate by at least 33 percent at 6 month horizon and 50 percent at a 1 year horizon.