Bayesian Analysis of Random Coefficient Dynamic AutoRegressive Model

Abstract

The goal of this work is to develop Random Coefficient AutoRegressive (RCAR) model and AutoRegressive (AR) model to Random Coefficient Dynamic AutoRegressive (RCDAR) model. The RCDAR model is considered by adding exogenous variables in RCAR model, so there are two variables in RCDAR model. This paper proposes the Bayesian analysis to estimate parameter of the first order RCDAR model. The noninformative prior is used to the Bayesian estimation procedure that works well for the AR model. Monte Carlo simulations was repeated for each situations in comparison of the coefficient from RCDAR model which is stationary, weakly stationary, and closed to nonstationary data. The results of coefficient estimator are satisfied weakly stationary data which is performed to fit possibly for large sample sizes.