The Analysis of Chinese Stock Volatile Risk Factors based on Mixture Distribution

An-xing WANG,Xian-ming TAN
DOI: https://doi.org/10.16381/j.cnki.issn1003-207x.2018.02.010
2018-01-01
Abstract:GARCH models have been widely used in modeling financial time series that exhibit time-varying volatility clustering.In this study,the model-based clustering approach is employed to examine clusters of 1165 stocks on Chinese security market on the basis of the estimated GARCH model parameters.It is found that the 1165 stocks could be divided into 4 clusters:cluster 1 consists of stocks with abnormal volatility features,while for stocks in the other three clusters.The distributions of the GARCH parameters have a similar shape but with different values.Stocks of manufacturing companies and decentralized nonstate-owned companies are more likely in the cluster with low volatility.Stocks of public utility companies (electricity,gas,water supply)and real-estate companies are more likely in the cluster with high volatility.
What problem does this paper attempt to address?