“…4 Firstly, our Bayesian estimation procedure, with the incorporation of the recent development of the MCMC method (Gelfand and Smith, 1990;Korteweg, 2012;Robert and Casella, 1999), is powerful and flexible in dealing with such a complex joint model, where the classical maximum likelihood approach encounters severe computational difficulties (Lopes and Carvalho, 2007). Finally, our approach allows us to 10 perform Bayesian model selection and cross-validation procedures, with considerable gains in computational efficiency over those used in conventional classical estimation approaches.…”