Application of Bayesian methods in the analysis of dynamic conditional correlation multivariate GARCH models

The study investigates the use and performance of the multivariate generalised autoregressive conditional heteroscedastic (MGARCH) model, specifically the dynamic conditional correlation (DCC)-MGARCH model in Bayesian framework. It uses a Markov chain Monte Carlo strategy and the Metropolis-Hastings algorithm for effective posterior sampling. The model is found to be more flexible and can describe uncertainties and volatilities of the error distribution. The sensitivity test shows that posterior results are more reliable when prior parameters are randomly sampled from the beta distribution.