Bayesian inference for term structure models

• Autor(a) principal: Martins, Thomas Correa e Silva
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://www.teses.usp.br/teses/disponiveis/104/104131/tde-09082022-085154/
• topic: Affine interest rate models; Asset pricing; Bayesian inference; Dynamic Nelson-Siegel; Inferência bayesiana; Modelos afins de taxas de juros; Modelos espaço de estados; Nelson-Siegel dinâmico; Precificação de ativos; State space time series

We explore recent advances in Bayesian methods in order to estimate the Vasicek, CIR and dynamic Nelson-Siegel (DNS) models for term structure of interest rates. The models are specified as state space time series. The main goal of this work is assessing and comparing the forecasting abilities of each model with respect to the observed data via mean absolute error. When estimated with synthetic simulated datasets, the models are able to successfully recover the latent vectors. As for the forecasting abilities, the multifactor models generally deliver the best predictions. The relevance of this work lies in integrating novel computational techniques for Bayesian inference with canonical models from the field of financial economics. Several aspects of both fields are discussed throughout the text.