Testing the Expectations Hypothesis of the Term Structure Using Instrumental Variables.

This paper demonstrates, by means of Monte Carlo experimentation, that tests of the expectations hypothesis of the term structure based on instrumental variables regressions of the change in the short rate on the relevant lagged yield spread are prone to severe over-rejection when the term premium is time-varying. In contrast, tests based on regressions of the yield spread on the first-difference of the short rate are found to reject at the correct rate in moderately sized samples. Copyright @ 1998 by John Wiley & Sons, Ltd. All rights reserved.