An improved Bayesian unit root test in stochastic volatility models

• Main Authors: LI, Yong, Jun YU
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2019
• Subjects: Bayes factor; Markov chain Monte Carlo; Posterior odds ratio; Stochastic volatility models; Unit root testing; Econometrics
• Online Access: https://ink.library.smu.edu.sg/soe_research/2311; https://ink.library.smu.edu.sg/context/soe_research/article/3310/viewcontent/aef200104_pv.pdf
• Institution: Singapore Management University

A new posterior odds analysis is developed to test for a unit root in volatilitydynamics in the context of stochastic volatility models. Our analysis extendsthe Bayesian unit root test of So and Li (1999) in two important ways. First,a mixed informative prior distribution with a random weight is introducedfor the Bayesian unit root testing in volatility. Second, a numerically morestable algorithm is introduced to compute Bayes factor, taking into accountthe special structure of the competing models. It can be shown that theapproach introduced overcomes the problem of the diverging “size” in themarginal likelihood approach by So and Li (1999) and improves the “power”of the unit root test. A simulation study is used to investigate the finite sampleperformance of the improved method and an empirical study implements theproposed method and the unit root hypothesis in volatility is rejected.