Bayesian Unit Root Tests in Stochastic Volatility Models

• Activity Number: 310
• Type: Contributed
• Date/Time: Wednesday, August 14, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Business & Economics Statistics Section*
• Author(s): Zeynep Kalaylioglu*+ and Sujit Ghosh and Sastry Pantula
• Affiliation(s): North Carolina State University and North Carolina State University and North Carolina State Unviersity
• Keywords: unit-root test ; gibbs sampling ; winBUGS ; stochastic volatility models ; markov chain monte carlo

Abstract

Persistence of shocks to volatility has important implications in finance; Business environment becomes uncertain and, as a result, markets plunge. This motivates the testing for a unit-root in Stochastic Volatility Models (SVM). Bayesian unit root tests based on the Bayes factors have been proposed in literature, using priors that are non-informative only on the stationary region. The support of such continuous densities does not include unity, the value being tested. This feature leads to poor frequentist performance of the Bayesian tests. We introduce a class of new prior densities that puts a positive mass at unity and develop Bayesian tests based on Bayes Factor and Posterior Intervals. Extensive simulation studies show that our proposed method performs much better in terms of reducing total (type I and type II) error rates, compared to other Bayesian methods and some frequentist methods. In addition, we develop a flexible software in WinBUGS that allows us to implement our method.