Font Size: 

This article discuss  the use of Bayesian methods for inference and forecasting in dynamic term structure models through Integrated Nested  Laplace Approximations  (INLA). This  method of analytical approximations  allows for accurate  inferences  for latent factors, parameters and forecasts in dynamic models with reduced computational cost. In the estimation of dynamic term structure models it also  avoids some simplifications in the inference procedures, as the estimation in  two stages.
The results  obtained in the estimation of the dynamic Nelson-Siegel model  indicate that this methodology performs  more accurate out-of-sample forecasts  compared to the methods of two-stage estimation by OLS and also Bayesian estimation methods using MCMC. These  analytical approaches also allow calculating efficiently measures of  model selection  such as generalized cross validation and marginal likelihood, that may be computationally prohibitive in MCMC estimations.

Terma Structure; Latent Factors; Bayesian Forecasting; Laplace Approximations.