Volatility correlation structure, dynamic network and portfolio implications of Chinese stock market

Abstract

This study combines complex networks and sliding window technology to construct a static and dynamic network of volatility of the stocks in CSI 300 index using the COVID-19 epidemic as an example to analyze the impact of public health emergencies on the correlation structure of stock volatility, as an extended application to mine low-risk stock portfolios that are more resistant to risks under the "mean-variance" framework. Research shows three implications. (i) During the outbreak period, the density of the stock market volatility network was significantly higher than before and after the outbreak, and the network structure was more intense during the outbreak period. The leading industries are the manufacturing and financial industries, and the source of market risk transmission comes from the key nodes of the two industries. (ii) The dynamic network shows that under the impact of the epidemic, the correlation structure of stock market volatility has undergone abrupt changes and the overall market risk is time-changing, which indicates that the sudden impact of degeneration breaks the original structure and triggers new information connections in the stock market. (iii) The degree of stock centrality affects investment portfolio returns, which means that core stock portfolios with greater network centrality during the relatively stable market period and the upward period perform better, and the peripheral stock portfolio has an advantage in the period when the market fluctuates due to sudden external shocks. Interestingly, peripheral stock portfolios with lower centrality are more resistant to risks under sudden shocks. The results of this paper can provide important enlightenment for stock market supervision and investment portfolio risk management.