Bayesian Inference of C-AR(1) time series model with structural break

Abstract

A variable may be affected by some associated variables which may influence the estimation and testing procedures and also not much important to model separately, such types of variables are called covariates. The present paper dealt the covariate autoregressive (C-AR(1)) time series model with structural break in mean and variance under Bayesian framework. Parameters of the model have been estimated considering appropriate prior assumptions and compared with maximum likelihood estimator. A simulation study has been carried out to validate the theoretical results, and then the same implemented on the monthly REER time series of SAARC countries. Both studies, empirical and simulation justify our findings. A unit root hypothesis is also tested for the model under study and gets satisfactory result. Keywords: Autoregressive Model; Bayesian Inference; Covariate; Structural break; Gibbs Sampler; Posterior Probability