The Survey-Based Term Premium in Comparison to the Realized Term Premium

This paper constructs a 10-year realized term premium from the 10-year zero coupon Treasury yield in year 0 and the ex post 3-month Treasury yields from years 0 to 10. The realized term premium swung wildly until the mid-1980's, and then fluctuated within a fairly stable range showing no trend. In comparison, the term premium derived from surveys of interest rate forecasts (survey-based term premium) was substantially lower than the realized term premium and trended downward since the early 1990s. The large and systematic forecast errors in combination with the stability of the realized term premium suggest possibilities that professional forecasters might have missed the term premium demanded by investors (ex ante term premium) by a wide margin and/or that investors forecast the future paths of interest rates more accurately than professional forecasters. It is also unclear that the survey-based term premium fairly represents the professional forecasters' estimate of the ex ante term premium, not to mention the ex ante term premium itself. While it would be a daunting task to verify these possibilities, it is fairly clear that surveys of interest rate forecasts are of limited value as an investment guide