We apply the Bayesian workflow approach by Gelman et al., 2020 of model building, inference and comparison to the modeling of the term structure of interest rates. We describe novel procedures of Bayesian data analysis, such as prior predictive checking, computational diagnostics, and posterior predictive checkings. We propose versions of the Vasicek, CIR and Diebold-Li models for the term structure, elicit prior distributions for the parameters, estimate the models with both simulated and real data, and compare the models. Our results show that the Diebold-Li models have better predictive capabilities than those of the affine (Vasicek and CIR) models, however our main goal is to illustrate how the Bayesian workflow approach is carried out in a practical setting of financial econometrics.
Comissão Organizadora
Anderson Odias da Silva
Claudia Yoshinaga
Ricardo D. Brito
Felipe Saraiva Iachan
Vinicius Augusto Brunassi Silva
Complete tool for submission, evaluation and publication of proceedings for scientific events.
