Abstract: Based on the jump diffusion model， jump intensity is assumed to follow the self-threshold autoregressive model (SETAR) to reflect the structural break of jump density， and the GARCH model is used to describe the diffusion process of asset price volatility. The SETAR-GARCH model is constructed by making the jump density control probability of jump dynamics in which volatility affects jump density and by making the GARCH model control diffusion process in which jump dynamics influences volatility. Taking Shanghai real estate index for example， the empirical study finds that Shanghai real estate index exerts the threshold effect and significant GARCH effect， with 35. 21% jump mutation probability. The total variance of asset returns is largely caused by extreme jump dynamics. Investors’ evaluation of historical volatility directly affects jump intensity expectation， and the interference from historical jump dynamics has exacerbated the current volatility in the diffusion process， which illustrates that volatility diffusion processes and jump dynamics have a two-way feedback effect.

Key words: jump diffusion, jump intensity, volatility diffusion, mixed GARCH model, feedback effect