The Research On Jump Dynamics In Chinese Stock Returns

As a new stock market, Chinese stock market lacks of operation mechanism, and with frequent intervention by Central government, stock price and its return show abnormal and sharp volatility, which is called jump dynamics.In order to study the jump dynamics of stock return, based on the research on Chinese stock return features, we develop a MRS-GARCH model, solving regime-switching problems as well as capturing leptokurtosis and fat-tailed characters. By separately analyzing the time-variation character and transition probability between two states, we conclude from empirical evidence that Shanghai composite index shows more fat-tailed character when it’s in the state of low jump frequency than the state of high jump frequency; and Shenzhen composite index is more unstable, and once high-jump-frequency state arrives, the probability of becoming buoyancy is much smaller than Shanghai composite index. At the same time, this is an indirect evidence of clustering effect of jumps.As for the detailed research on jump dynamics, this paper introduces ARVI-GARCH-Jump model constructed by Chen&Sun (2010), the jump intensity of which expressed by autoregressive construction. In the model, we consider the time-variation, clustering effect, and jump feedback in the GARCH component and volatility feedback in the jump component. The empirical evidence shows that jump dynamics indeed exist in Shanghai and Shenzhen stock market. The average proportion of jump innovation variance of stock return in Shanghai and Shenzhen stock market to total conditional variance is57.8%、59.1%separately, and the jump size of Shenzhen stock return is bigger than Shanghai stock return. The jump dynamics of two stock returns show obviously clustering effect and time-variation character. And there are direct and interactive feedback effects between jump dynamics and conditional volatility. The traditional asymmetry in stock returns is obviously weakened by the jump component in news innovation. There exists shift effect in conditional volatility in two stock markets, and investors have return expectation with an unknown innovation impact.After capturing the behavior of jump with ARVI-GARCH-Jump model, we apply the non-parameter test to do model specification test, which shows that the ARVI-GARCH-Jump model can capture the distribution features of Shanghai stock return better than that of Shenzhen stock return.In addition, this paper discusses the effects of the jump on assets pricing theory and so on, for example, jump dynamics can make the risk hedging strategy hard to implement; however, jump dynamics provide opportunities for investment (or more should be called a "speculation").