19 Pages Posted: 17 Feb 2006
Date Written: October 16, 2002
One of the most popular theories of the term structure of interest rates is the pure expectations theory. In its purest form it postulates that long term interest rates are a reflection of the market's perception of future short term rates. Almost all of the tests of the pure form of this theory with financial markets of various countries have resulted in a rejection of the theory. A standard approach of most of the literature is to linearize the equations associated with the theory and then estimate a linear regression to determine if the spread is indeed a predictor of future short rates. While this approximation is valid for testing the expectations hypothesis when the spread is between long and short duration instruments, researchers should be careful applying this technique to money market spreads. In particular, this approximation will not be valid for very flat term structures, which is usually the case for bills with less than twelve months to maturity.
This paper uses simulations to show how erroneous conclusions can be drawn by using this approximation in the money market and re-examines the pure expectations theory for the money markets of various countries using the DOLS approach to estimating series that are cointegrated. The conclusions are different from other studies on this subject and are particularily supportive of the expectation's hypothesis. In fact, while the theory is sometimes rejected (accepted) when using approximations, it is accepted (rejected) using the exact formulas.
Keywords: money markets, expectations theory of term structure, fixed income
Suggested Citation: Suggested Citation
By N. Mankiw