• preprint
    • Published in RePEc
Abstract
Controversies in term-structure theory center around the existence and variability of term premia in securities yields. In this paper, the term premium on a default-free n-period bond is defined as the difference between its observable yield to maturity and the average expected per-annum rate of return on an n-period strip of rollover investments in one-period bonds. To test alternative term-structure theories without introducing ex post proxies for expectational variables, this paper uses a set of cross-section interest-rate forecasts collected jointly with Burton Malkiel of Princeton University from a population of large institutional lenders at four different phases of a single interest-rate cycle. Statistical tests strongly confirm the existence of nonzero term premia at each survey date, thereby rejecting the pure-expectations theory of the term structure. Additional tests are unable to reject restrictions implied by the liquidity-premium hypothesis that term premia should be positive and increase with maturity. Finally, contrary to the martingale hypothesis, ex ante term-premium data vary significantly overtime and show a positive association with the level of interest rates.
All Related Versions

This publication has 0 references indexed in Scilit: