Forecasts of the Canada-U.S. exchange rate: efficient markets versus purchasing power parity

Business Economics, Oct, 1991 by Peter S. Spiro

WIDESPREAD, and justified, perception prevails that economists' forecasts of exchange rates are even more error prone than other types of economic forecasts.(1) The present article will evaluate this issue by examining the actual record of Canadian dollar forecasts in a period of unusual turbulence. It will attempt to explain the sources of error in economists' forecasts and try to determine to what extent, if any, these forecasts can be improved.

A well-known study by Meese and Rogoff (1983) evaluated a number of economic models of exchange rates for their out-of-sample forecasting ability. They concluded that these models could not outperform the random walk hypothesis. To some extent, this may be a problem of the inadequacy of the models assessed by Meese and Rogoff, which were the popular and highly theoretical models of the 1970s. In any case, Meese and Rogoff focused on artificial forecasts generated by models, rather than the actual forecasts the economists make using their judgment. It will be interesting to evaluate how economists' actual forecasts of the Canadian dollar compare against the random walk.

THE EFFICIENT MARKETS HYPOTHESIS

According to one school of thought, the EFficient Markets Hypothesis (EMH) implies that it is futile for economists to try to forecast exchange rates. An exchange rate is determined in a free auction market, with a large number of traders bidding against each other in an attempt to maximize profits. These traders have a great incentive to utilize fully all available information. Therefore, the value at which the currency trades at any instant in time is necessarily the best forecasts of where it will be tomorrow: if there were a good reason to believe that it was going to rise further tomorrow, traders would already have bid its price up to that level today. Therefore, EMH means that the price of a financial asset will be unpredictable. The future price will follow a random walk as unforeseen new pieces of information arise that cause the market to reconsider its value.

In spite of the popularity of the EMH, one cannot take its reliability for granted without checking it further. A large body of empirical evidence in favor of the EMH is available in the finance literature, but it tends to be focused on the stock and bond market and on very short time horizons. The work of Robert Shiller (1989) has shown convincingly that short-term market efficiency, of the type implied by the EMH, is not necessarily the same as full-fledged rationality. A market may be efficient, in the sense that it is unpredictable, while failing to bring about the correct value that achieves a balance between supply and demand.

The volume of foreign exchange trading is phenomenal. Froot and Thaler (1990, 180) report that in 1989 it amounted to $430 billion per day, about forty times greater than the total daily world trade in goods and services of $11 billion. Clearly, trading in these volumes can result in exchange rates that are determined by trading activity itself and have nothing to do with economic fundamentals. However, if an exchange rate deviates far from its fundamental value, the trade balance will be affected, and this will set off forces that eventually push the exchange rate back to the rational market equilibrium. Therefore, it is valied to ask whether economists can outforecasts the market; the answer is likely to be that it is a matter of time horizon over which forecasts are made.

THE EFFICIENT MARKETS HYPOTHESIS AND

THE FORWARD EXCHANGE RATE

In the case of exchange rates, a serious complication exists in the application of the EMH. At one time, it was assumed that the forward exchange rate represented the market's expectation of the actual future value of the exchange rate. However, it is now realized that this need not be the case if risk aversion is a significant factor limiting international capital flows. Alternative theories about the causes of the bias in the forward rate are surveyed by Froot and Thaler (1990).

A necessary arithmetic relation exists between the forward rate and the interest differential. For example, if the Canadian interest rate is 1 percentage point higher than its U.S. counterpart, then the one year forward rate for the Canadian dollar must represent a 1 percent depreciation of the Canadian dollar vis-a-vis the U.S. dollar. This is called covered interest arbitrage, and if it did not hold then an opportunity would be available to make a riskless excess profit by lending in one country or the other. The forward rate is the mirror image of the interest rate differential that prevails between Canada and the United States, and in principle either one of the could be the cause of the other. If the forward rate were the market's actual expectation of the future value of the exchange rate, it would imply that interest rate differentials are determined by the expected change in the exchange rate embodied in the forward rate. In that case, Canada could have a higher interest rate than the United States only because the market expected the Canadian dollar to decline correspondingly in the coming year.