A Study On The Volatility Spillover Effect Between Stock Markets Based On Beyesian Heavy-tailed DCC-MSV Model

Since1980s, with the development of economic globalization and financialintegration in the global scope, the interdependence between financial markets aroundthe world has been greatly strengthened. The volatility of a single financial market isnot only influenced by its own previous fluctuations, but also affected by the previousfluctuations in other financial markets, that is to say, there exists the volatilit yspillover effect. The existing research on financial market volatility spillovers mostlyused GARCH class model, which is different from GARCH class models, stochasticvolatility model is is considered more suitable for the empirical study of the financalsector. Therefore, this paper proposes a Bayesian heavy-tailed DCC-MSV model withbidirectional Grainger test, to research t he volatility spillover and dynamic correlationbetween different financial markets.First, in terms of the Bayesian analysis of heavy-tailed DCC-MSV model, thepaper constructs a Markov Chain Monte Carlo algorithm procedure to estimateparameters, avoiding the difficulty of parameter estimation. Then, this paperintroduces the convergence diagnostics method of the stochastic volatility model’sparameters, the comparison criteria of stochastic volatility model and the significanttest method of volatility spillover effect. Finally, we select the CSI300Index, theHang Seng Index, the S&P500index, the FTSE100index, the Nikkei225index as theobject of study, to analyze the volatility spillover effect and the dynamic relationshipbetween the stock markets in China, Hongkong, U.S., UK, Japan after the stockmarkets of China implemented the shareholder structure reform in2005. At the sametime, we use DIC criteria to compare CCC-MSV model, heavy-tailed CCC-MSV model,DCC-MSV model and heavy-tailed DCC-MSV model.The results show that in terms of the volatility spillover, only the stock markets ofHong Kong have unidirectional volatility spillover effect on the stock markets ofChina. What’s more, the stock markets of U.S., the stock markets of UK and the stockmarkets of Japan don’t have volatility spillover effect on the stock markets of China.And we don’t find volatility spillover effect from the stock markets of China to theother four stock markets. However, it is also found that the bidirectional volatilit yspillover effect among the stock markets of Hong Kong and UK, the stock markets ofHong Kong and Japan, the stock markets of U.S. and UK. Further, we find unidirectional volatility spillover effect from the stock markets of U.S. to both thestock markets of Hong Kong and Japan, and it is also found that the unidirectionalvolatility spillover effect from the stock markets of UK to the stock markets of Japan.In terms of the dynamic correlation, the correlation between stock markets hastime-varying characteristics. In addition, the correlation has long memory. At the sametime, we find that the correlation shows a rising trend during the financial crisis.Finally, the correlation between the stock markets of China and the stock markets ofHong Kong is closer. In terms of the simulation effect of volatility model, heavy-tailedCCC-MSV model and heavy-tailed DCC-MSV model are better than CCC-MSV modeland DCC-MSV model.