Systematically discovering dependence structure of global stock markets using dynamic Bayesian network

This paper aims to investigate the dependence structure of global financial markets using an systematic analytical framework of Dynamic Bayesian Network (DBN). DBN, an temporal extension of Bayesian Network, admits contemporaneous and lagged nonlinear conditional dependencies among markets without specifying functional forms like copula. Therefore, DBN can tell not only how global stock markets interact in one calendar day but also how the returns in previous day in?uence the present market performance. Several elementary characteristics of the dependence among markets, such as the evolving property and asymmetric dependence, can also be well captured and analyzed in our analytical framework of DBN. The computational results demonstrate the feasibility and effectiveness of the proposed method and some important conclusions are obtained. ? 2013 Binary Information Press.EI0187215-7226