On building causal networks for Chinese stock market understanding

Abstract

This study proposes a causal network construction method based on Granger causality test rather than correlation coefficient to investigate the inherent structure of the stock market. We analyze the characteristics, community structure and nodes' influence of the network formed following our method. Furthermore, this study investigates the reasons why the stock market performs a certain relationship among stocks and why some stocks can be important in the stock market. This is the first paper tries to illustrate the mechanism of stock network. This paper proposes a new way to analyze the formation reasons of community structure and important nodes of the stock market, which integrates complex networks and financial econometric methods. This paper conducts empirical study for Chinese stock market to illustrate the usefulness and advantage of our new methods. The empirical results show that it is the pricing factors such as yearly abnormal returns, price volatility, price level and leverage that drives the stocks rather than industry sector. Our study provides new evidence to help us understand the stock market and stock pricing. Particularly, the results of this paper can help understand the sector rotation effect in the stock market.