What you need to know about exchange rate forecasts
There is no reliable method available to forecast exchange
rates. Paul Krugman and Maurice Obstfeld write in their book "International
Economics: Theory and Policy" (New York, HarperCollins:1994):
If exchange rates are asset prices that respond immediately to
changes in expectations and interest rates, they should have properties
similar to those of other asset prices, for example, stock prices. Like
stock prices, exchange rates should respond strongly to "news," that is,
unexpected economic and political events; and, like stock prices, they
therefore should be very hard to forecast. In a study by Richard M.
Levich ("Evaluating the Performance of the Forecasters", in: Donald
R. Lessard, ed., "International Financial Management", New York: John
Wiley:1985, pp. 218-233) in which he surveyed the track record of
a dozen exchange rate forecasting companies through 1982 he found
little evidence that professional forecaster do systematically better
than an individual who, for example, uses the three month forward rate.
That does not imply that forward rates are good predictors. In fact,
they usually do not predict exchange rate movements very well.
How are exchange rates determined?
Long term movements are driven by fundamental forces, such
as purchasing power parity. In the short run,
exchange rate movements are driven by news and events.
For excellent surveys of exchange rate theory see:
• Richard Baillie and Patrick C. McMahon:
The foreign exchange market: theory and econometric evidence.
Cambridge University Press, New York, 1989.
• Lawrence S. Copeland: Exchange Rates and International Finance.
Addison-Wesley, Reading/Mass., 1990.
• Christian Dunis and Michael Feeny (editors): Exchange Rate
Forecasting. Woodhead-Faulkner, New York, 1989.
• Peter Isard: Exchange Rate Economics. Cambridge University
Press, Cambridge, 1995.
• Ronald MacDonald and Mark P. Taylor: Exchange Rate Economics:
A Survey. IMF Staff Papers, 39(1), 1-57.
If exchange rate forecasts are not reliable, why bother?
As pointed out above, economists do not possess reliable methods
of forecasting exchange rates over short time horizons such as
days or weeks. Fundamental economic forces (such as purchasing
power parity and balance-of-payment disequilibrium) usually take
much longer, often several years, to have an effect on exchange
rates. Nevertheless, the business community has to form
expectations about the short-term trends of exchange rates. Because
these expectations are formed by market sentiment, trends can be
extrapolated. At the same time, market sentiments can change as
financial news can change the outlook on a country's economic
prospects. The PACIFIC Exchange Rate Service provides two sets of
exchange rates trends using two different "technical" trending
techniques for 30-day and 60-day intervals.
Trends based on economic fundamentals (usually for quarterly
or annual projections) are usually obtained through large structural
computational models, such as the FOCUS model of
and Economic Analysis Program at the University of Toronto.
The PACIFIC trends are intended to reveal market trends, not
fundamental determinants. Users of the trend charts are advised
to read them with considerable caution.
How the trends were calculated
The diagrams below show 60-day trends and 30-day trends for the
Canadian Dollar vis-à-vis four major currencies: the US
Dollar, the German Mark, the Japanese Yen, and the British
Pound. Understand that these trends are not
forecasts. They do not predict the future; they only
visualize current trends that can be broken at any time due to the
arrival of new information. |
In the 60-day trend
diagrams, the solid blue line describes the trended path of the
exchange rate, and the light blue area describes the 95%
confidence interval. The method used in the trends is a combined
quadratic trend model and autoregressive model, called a
stepwise autoregressive method. The trend component is
supposed to capture long-term behaviour, while the autoregressive
component is supposed to capture short-term behaviour. The
autoregressive parameters included in the model for each series
are selected by a stepwise regression procedure, so that
autoregressive parameters are only included at those lags at which
they are statistically significant. Since the trend and
autoregressive parameters are computed in sequence rather than
simultaneously, there are objections against this method on a
theoretical level. However, the trends are computationally
inexpensive. The computations are carried out using
PROC TREND in the SAS
The 30-day trend diagrams are based
on "runs" of exchange rate movements within a tunnel. This is
essentially a heuristic trending technique. The tunnel is
constructed to be as long and as narrow as possible. The tradeoff
between length and width is determined by a scoring function
defined by the trender. The computations are carried out using
PROC IML in the SAS
language with code developed by the author.
I make no warranties, express or implied, that the exchange rate
trends/forecasts are accurate, useful, consistent with any economic
theory, reliable, or consistent with any particular standard of
merchantability, or that they will meet your requirements for any
particular application. These forecasts should not be relied upon for
carrying out any currency transactions which could result in injury to
a person or loss of property. If you do use the forecasts in such a
manner, it is at your own risk. I disclaim all liability for direct
or consequential damages resulting from your use of these forecasts.