A Recursive Parameter Estimation Technique for Term Structure Models

Abstract

In this article, the authors develop a new method of estimating multi-parameter term structure models using panel data. This technique involves recursively estimating some parameters along the cross-sectional dimension and the rest of the parameters along the time series dimension until convergence is achieved. By breaking down the parameter estimation process into two simpler procedures along these dimensions, the authors are able to isolate and solve common problems plaguing other methods such as quasi-maximum likelihood estimation via the Kalman filter. As a demonstration, they apply this technique successfully to the one-factor Vasicek and two-factor Cox–Ingersoll–Ross models using Fama–Bliss Treasury data. Simulation results indicate that this technique yields reasonable and robust parameter estimates for these models.