Abstract: Recently, the unexpected events, such as the US financial crisis, the EU debt crisis and the earthquakes, constantly shock China's financial markets. Under these impacts, almost all sorts of financial assets suffer frequent jumps in their prices, which then cause much higher risk in the short period of time. In order to describe both of the normal volatility and the jump change in financial asset, a mixed jump-GARCH model,that is TSD-ARJI-GARCH model,is constructed in this paper. It extends the existing models to make the jump intensity not only driven by internal idiosyncratic factors, but impacted by external state variables. This model takes into account time variation, clustering and threshold-based state-dependence of jump change as well as asymmetry and clustering of normal volatility. Then, stock price data of listed companies in China's stock market is used to test these different models. Both goodness of fit and forecast evaluation show TSD-ARJI-GARCH model performs better. Therefore, it can provide relatively better theoretical and technological support for monitoring and managing the jump-risk of financial assets in practice.

Key words: asset price, jump, risk, variable intensity, threshold effect