A BAYESIAN INFERENCE OF MULTIPLE STRUCTURAL BREAKS IN MEAN AND ERROR VARIANCE IN PANEL AR (1) MODEL
A BAYESIAN INFERENCE OF MULTIPLE STRUCTURAL BREAKS IN MEAN AND ERROR VARIANCE IN PANEL AR (1) MODEL

Summary/Abstract: This paper explores the effect of multiple structural breaks to estimate the parameters and test the unit root hypothesis in panel data time series model under Bayesian perspective. These breaks are present in both mean and error variance at the same time point. We obtain Bayes estimates for different loss function using conditional posterior distribution, which is not coming in a closed form, and this is approximately explained by Gibbs sampling. For hypothesis testing, posterior odds ratio is calculated and solved via Monte Carlo Integration. The proposed methodology is illustrated with numerical examples.

• Details
• Contents