关键词: 股市泡沫风险, 逻辑平滑转换自回归模型, 蒙特卡洛模拟, 检验功效

Abstract: This paper fits the real data generation process of the main stock price indexes in China’s stocks market under the nonlinear model frame, and proposes a method of stock price bubbles to pre-judgment the stock price bubble risk and provide reference for policy measures to prevent financial risks, which is more effective test than the supremum unit root (SADF) method proposed by Phillips et al. (in 2011). Empirical test finds that major stock price indexes all present logistic smooth transition autoregressive characteristics. Specifically, there have been bubble risk existing in the four indexes since their launch, with six main persistent periods for Shanghai securities composite index, four main persistent periods for Shenzhen component index, two main persistent periods for Shanghai and Shenzhen 300 index and three main persistent periods for growth enterprise index. Overall, the bubble generation time of growth enterprise index will be ahead of the other three indexes, which can be taken as the leading indicator for the early warning of China’s stock market price bubble risk. There is no bubble risk in China's stock market after July 2015.

Key words: Stock market bubble risk, Logistic smooth transition autoregressive model, Monte Carlo simulation, Test power.