Forecasting the Term Structure of Interest Rates Using Integrated Nested Laplace Approximations

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.