Time-Consistent No-Arbitrage Models of the Term Structure
48 Pages Posted: 27 Feb 2001
Date Written: February 2001
We develop and empirically explore a model of the term structure that captures as well as possible the time-series dynamics of a set of state variables and fits exactly the date-to-date cross-sections of bond prices. We construct our model in two stages. In the first stage we use a flexible time-series model to describe the dynamics of the state variables. In the second stage, we infer for each date in the sample the pricing kernel from the term structure of interest rates using conditional moments implied by the restriction of no-arbitrage. We estimate a number of two-factor models for U.S. term structure data and use various in- and out-of-sample measures to evaluate the relative importance of well-specified factor dynamics and non-linearities in the pricing kernel. We find that both aspects are crucial for out-of-sample pricing. Specifically, our results lend support to quadratic term structure models.
Suggested Citation: Suggested Citation