Visit Link

Description:
by C. Alan Bester

Traditional affine models of the term structure are eminently tractable, but suffer from empirical difficulties. Random field models offer great flexibility in fitting the data, but are widely considered non-implementable unless they are approximated by a low-dimensional system. I develop a state-space estimation framework where both random field and affine models can be estimated by MCMC using the same panel of forward rate data. I find that random field models are much better able to fit the patterns of volatility and correlation in a long historical sample of U.S. Treasury forward rates.
Discuss this paper

(No registration is required)

   Visit Link

Description:
Francis A. Longstaff*, Pedro Santa-Clara,
Eduardo S. Schwartz
Abstract This paper studies the costs of applying single-factor exercise strategies to American swap options when the term structure is actually driven by multiple factors. Using a multifactor string market model of the term structure, we find that even when singlefactor models are recalibrated to match the market at every exercise date, the exercise strategies they imply can be suboptimal. Based on estimates of notional amounts outstanding, the total present value costs of following single-factor strategies could be several billion dollars. These results illustrate the importance of using well-specified term structure models. r 2001 Published by Elsevier Science S.A
Discuss this paper

(No registration is required)