Modelo de volatilidade estocástica com mudanças de níveis aleatórias: uma aplicação aos índices do mercado de ações latino-americano

Abstract

In this work, we study a stochastic volatility model with random level shifts proposed by Qu and Perron (2013) to analyze financial return series from a Bayesian perspective. The estimation of log-volatilities and the parameters of interest was performed using MCMC methods and data augmentation. As an application of the studied techniques, we conducted examples with simulated data and an empirical study of five Latin American indices. Additionally, we compared the results obtained with the basic volatility model with normal errors and used the DIC comparison criterion for model selection.