Bayesian Estimation for Fully Shifted Panel AR(1) Time Series Model

Jitendra Kumar, Varun Agiwal

Abstract

Present manuscript proposed a shifted panel autoregressive (PAR) model through structural break assumption. A Bayesian estimation method is developed considering known from of prior information. Since expression of posterior distribution under different loss functions is in complicated form, therefore Gibbs sampler technique is used to obtain the conditional posterior distribution. A simulation and empirical study for proposed shifted panel AR(1) model is carried out to record the performance of the Bayes estimators and compared with the classical procedures such as maximum likelihood and least square estimator. A realization of real data set is also explored to illustrate the prospective interpretation of the proposed model.