ince the level of interest rates is so critical to the profitability of many busi-S ness organizations, particularly financial institutions, much time and effort is spent attempting to forecast their probable future course. This activity has become even more critical with today's highly volatile markets.
Currently there is a wealth of economic and financial data from which interest rate forecasts can be derived. The newest and easiest method is to use the interest rate quoted for one of the interest rate futures contracts as a forecast of future rates. The primary purpose of this article is to test the forecasting accuracy of one of these interest rate futures contracts: the 13-week Treasury-bill futures contract traded on the Chicago Mercantile Exchange. We compare the forecasting accuracy of the 13-week T-bill futures contract with the closely related forward rate implied by the term structure of interest rates and to two naive forecasting models. We conclude that the T-bill futures rate does provide as good a forecast as do the other methods, although its accuracy is far from perfect.
We start our discussion by reviewing the wide array of articles that have been published in the area of interest rate forecasting. We then turn to testing the forecast accuracy of the T-bill futures rate and examine the relationship between the futures rate and the forward rate.
I. REVIEW OF THE EVIDENCE
The largest body of academic literature dealing with the forecasting of interest rates explores the predictive power of the forward rate implied by the term structure of interest rates. One theory of the term structure of interest rates, the so- called pure expectations theory, postulates that long-term rates are simply a geometric average of the intervening expected short-term rates. A forecast of future interest rates can be obtained by "unwinding" the yield curve by using the appropriate combination of a long-term rate and a short-term rate to obtain the forward rate. If the pure expectations theory is valid, it is reasoned, the forward rate should be a good forecaster of the future interest rate. Therefore, a number of studies have been undertaken to check the forecasting accuracy of the forward rate. The general conclusion of these studies is that the forward rate, because of the existence of a variable liquidity premium, is a poor forecaster of future in-