Abstract: To appropriately measure the overall contagion effect in different financial markets, a multivariate stochastic contagion model according to multivariate GARCH model including the mean spillover equation, the fluctuation spill equation and the contagion equation is proposed. The new contagion equation is proposed by combining the real financial contagion phenomenon which is different from the fixed coefficient contagion variable and the general covariance contagion variable. Then, a whole solution method based on the MCMC iterative algorithm is given out. In the end, the contagion effect in copper and other metal futures markets using the new contagion model are fitted and forecasted. The empirical results verify some conclusions from previous literatures and give out some new conclusions such as the contagion accumulative effect, the risk alternately phenomenon between the copper and aluminum futures markets, the slow market change in the contagion phenomenon and so on. Moreover, the empirical result also shows the effectiveness, practicality and superiority of this new contagion model.

Key words: mean spillover, fluctuations spillover, contagion, MCMC iterative algorithm, copper futures