Are Dollar Exchange Rates Cointegrated After All?

Abstract

Movements in the exchange rate of the dollar against the major currencies have occupied the centre stage on foreign exchange markets and in the preoccupations of specialists of international finance ever since the beginning of the floating exchange rate period. However, it is the appreciation of the dollar in the first half of the 1980s which has aroused special interest. Indeed, over such a period — even though this would go against the assumption of weak efficiency — one is tempted to search for a common trend in the exchange rates of major currencies vis-a-vis the dollar. This indeed has been the object of a lively controversy over the last few years. Baillie and Bollerslev (1989) opened the debate, using the two-step Engle-Granger technique and the more powerful Johansen test (see also chapter 4), by showing that a unique cointegrating relationship exists between the spot dollar exchange rates of the G6 currencies plus the Swiss franc over March 1980–January 1985 using daily data. However, the ‘fragility’ of such a result was evidenced by Sephton and Larsen (1991), who argued that the Johansen test exhibits a strong sensitivity to the time period on which it is based; that is, the choice of the sample period is crucial in determining the existence (or absence) of cointegration.