Purpose: To propose a Bayesian testing procedure for unit-root in stochastic volatility models with correlated errors. Summary: It is observed that the generalized autoregressive conditional heteroskedasticity (CARCH) and the stochastic volatility (SV) models are used to capture the time evolving feature of volatility. In fact, a series of returns are often modeled using SV models. Since many observed series exhibit unit-root non-stationary behavior in the latent AR(1) volatility process, it is observed that tests for a unit-root is necessary, particularly when the error process of the returns is correlated with the error terms of the AR(1) process. Therefore, in this article, a Bayesian testing procedure for unit-root in SV model is proposed. In this regard, a suitable class of priors that assigns positive prior probability on the non-stationary region is considered. Then the posterior credible interval is employed for the hypothesis testing decision criterion. An extensive simulation study is conducted to demonstrate the performance of the proposed testing method. Results of an empirical study is presented in support of this. (24 refs.)
展开▼
