ABSTRACT
Prior studies indicate that the predictive power of implied forward rates for future spot rates is weak over long sample periods and typically varies dramatically across different subperiods. Fama (1976, 1984) conjectures that the low forecast power is due to a failure to control for the term premium embedded in forward rates. We show that Fama's conjecture is consistent with the data using any of four different models of the term premium. We measure the term premium using a variety of ex ante instruments, including the junk bond premium, bid-ask spreads in Treasury bills, the Standard & Poor's 500 stock index's dividend yield and the conditional volatility of interest rate changes using an Autoregressive Conditionally Heteroscedastic (ARCH) process. Forward rates adjusted for the term premium are reliable predictors of future spot rates over the entire 1963-1993 period.

Download the paper (Acrobat .pdf file) . Download Acrobat Reader.