Unit Root Testing for the Time Series Model with Additive Outlier: A Bayesian Approach

Abstract

The presence of outliers in time series may have serious implications in the estimation of parameters or testing of hypothesis. The analysis of time series in the presence of outlier is not much explored using Bayesian framework. The present paper deals with the Bayesian analysis of an autoregressive model involving linear time trend and contaminated by additive outlier. The issue of unit root hypothesis is dealt in Bayesian set-up and posterior odds ratio for the unit root hypothesis has been derived under appropriate prior assumptions. The numerical illustration is carried out to observe the impact of presence of an additive outlier on posterior odds ratio for unit root hypothesis.