This report presents different tests of market efficiency in the when- issued market for Government of Canada treasury bills and examines the effectiveness, in this market, of Bank operations over the 1986 to mid- 1992 period. The when-issued market, which is a combination of a forward and futures market, enables market participants to buy or sell treasury bills up to one week in advance of the regular weekly auction. The theoretical pricing relations presented show that forward and futures yields, even in informationally efficient forward markets, may not be unbiased estimates of future spot yields. This occurs because a sweetener may be necessary to entice one side of the market into covering the other and because cash flows are uncertain at the time investment decisions are made.

In general, the ordinary least squares regressions reveal only marginal justification for rejecting the composite null hypotheses, which include market efficiency and rational expectations. However, the report shows that when efficiency gains are exploited, through modelling of the time-varying volatility, as generalized autoregressive conditional heteroscedasticity (GARCH), yield changes tend to be followed by yield changes of the same sign for the three when-issued bills. The empirical evidence also suggests that unexploited profitable hedging opportunities may have existed in the three- and six- month coincident-to-when-issued bills. Finally, the report shows that, when the null hypothesis is rejected, when-issued yields understate treasury bill yields at auction and that open market operations influence when-issued yields and auction yields differently.