Variance Persistence in the Greater China Region: A Multivariate GARCH Approach

Abstract

This paper utilizes three Multivariate General Autoregressive Conditional Heteroscedasticity (MGARCH) models to determine variance persistence in the Greater China region from 2009 to 2014. The first approach applies the Baba, Engle, Kraft and Kroner (BEKK) model and shows that the Shanghai Stock Exchange Composite Index (SSEI), Taiwan Capitalization Weighted Stock Index (TAEIX) and the Hang Seng Stock Index (HSEI) stock returns are all functions of their lagged covariances and lagged cross-product innovations. The second MGARCH approach applies two methodologies, namely, dynamic conditional correlation (DCC), and constant conditional correlation (CCC) estimations. The DCC model concludes both short- and long-run persistencies between Taiwan’s TAIEX and Hong Kong’s HSEI. Alternatively, the CCC model confirms the initial findings of the BEKK model, and adds that the relationships among these three strong economies are stable in the long-run. The log-likelihood values determine that the DCC model is better in judging volatility dynamics in the Greater China region, because of economic clauses brought by the Closer Economic Partnership Arrangement (CEPA), the Economic Co-operation Framework Agreement (ECFA) and the Hong Kong - Taiwan Business Cooperation Committee (BCC).