Priors and Bayesian parameter estimation of affine term structure models

Affine term structure models describe the term structure of interest rates by means of a small number of latent factors. Quasi-unit root behaviour for these latent factors arises from the high degree of serial correlation in interest rate data. In this paper we perform Bayesian parameter estimation and demonstrate that the close to unit root behaviour of the latent factors should be considered properly. We show that with increasing serial correlation the Fisher information matrix approaches a singularity. We apply Markov Chain Monte Carlo simulation techniques in conjunction with regularised priors to simulate the joint posterior distribution of the model parameters. Informative priors are necessary to obtain a well performing Bayesian sampler.