The Efficiency of the Treasury Bill Futures Market

On January 6, 1976, the International Monetary Market of the Chicago Mercantile Exchange began trading the Treasury bill futures contract. Compared with other financial instruments, the Treasury bill futures contract is relatively simple to price. As we will show, the equilibrium price of a contract can be determined or closely approximated from observable spot Treasury bill prices. Given the ease of pricing the contract, one would expect the market for Treasury bill futures to be highly efficient. The purpose of this paper is to test the efficiency of the Treasury bill futures market. The specifications of the contract call for delivery of a $1,000,000 par value, 90day U.S. Treasury bill, although 91 or 92 day bills are substitutable. The delivery months are March, June, September, and December, and eight contract maturities are currently traded. Contracts are deliverable on the second day following the Federal Reserve 3-month (13 week) Treasury bill auction of the third week of the delivery month. This generally falls on the third Thursday of the month. Recently, several studies (Branch [2], Capozza and Cornell [3], Emery and Scott [5], Lang and Rasche [6], Oldfield [7], Poole [8], and Puglisi [9]) of the Treasury bill futures market have attempted either to test the efficiency of the market or to test the empirical validity of the Expectations Hypothesis of the term structure of interest rates. Both types of tests focused on the existence of arbitrage opportunities between the futures and spot markets. Although the nonexistence of arbitrage opportunities does not prove the Expectations Hypothesis, the studies that intended to test the Hypothesis can be viewed as indirect tests of market efficiency.' These studies find conflicting evidence regarding the efficiency of the futures market. These conflicting results can be partially explained by differences in sample sizes and sample periods. However, it is possible that the same data may not have been employed in a consistent manner in all studies. For example, only in the Branch, Poole, and Lang and Rasche papers is there any recognition of the fact that the delivery vehicle for contracts, other than the nearby contract, has never existed. Therefore, it is not clear how this problem was resolved in the development of the data to be tested. In addition, several studies recognize the