may therefore deviate markedly from the expectation embodied in the present for-
ward rate for that maturity.
As is indicated in Exhibit 6.8, in an efficient market the forecasting error will be
distributed randomly, according to some probability distribution, with a mean equal
to zero. An implication of this is that today’s forecast, as represented by the forward
rate, is equal to yesterday’s forward plus some random amount. In other words, the
forward rate itself follows a random walk.^1
Another way of looking at these is to consider them as speculative profits or
losses: what you would gain or lose if you consistently bet against the forward rate.
Can they be consistently positive or negative? A priori reasoning suggests that this
should not be the case. Otherwise, one would have to explain why consistent losers
do not quit the market, or why consistent winners are not imitated by others or do
not increase their volume of activity, thus causing adjustment of the forward rate in
the direction of their expectation. Barring such explanation, one would expect that
the forecast error is sometimes positive, sometimes negative, alternating in a ran-
dom fashion, driven by unexpected events in the economic and political environ-
ment.
Rigorously tested academic models have cast doubt on the pure unbiased forward
rate theory of efficiency, and demonstrated the presence of speculative profit oppor-
6.6 GUIDELINES FOR CORPORATE FORECASTING OF EXCHANGE RATES 6 • 21Exhibit 6.8. The Unbiased Forward Rate Theory. This theory says, in effect, that the forward
rate follows a random walk; this implies that the spot rate follows a random walk with drift.
For wardSpotToday In three
monthsTIMEActualProbability
distribution
of actual
exchange rateEXCHANGE RATE(^1) Note that when we say the forward rate follows a random walk, we mean the forward for a given de-
livery date, not the rolling three-month forward. Since the only published measure of a forward rate for
a given delivery date is the price of a futures contract, the latter serves as a proxy to test the proposition
that the forward rate should fluctuate randomly.