Bayesian Inference for C-AR Model with Spherically Symmetric Error

Abstract

Present work has investigated the inferences in covariate autoregressive (C-AR) model when error term is non-Gaussian. Spherically symmetric behaviour is one of the non-Gaussian characteristic and may be seen in agricultural, economics and biological field. So, we have considered here the spherical symmetric error in C-AR model. Under Bayesian methodology, estimators of the model parameters are obtained by conditional posterior distribution. A Bayes factor is derived for testing the unit root hypothesis. A simulation study has conducted when error term follows t-distribution. In empirical study, we considered REER time series of SAARC countries to illustrate the applicability of the proposed model.