A Four-Factor Model with Characteristics of Stock Correlation Network : Evidence from China

We find effectively extracting information from the stock-related structure (SRS) can improve the predictability of stock returns. A new factor for asset pricing is constructed that captures the characteristics of stock correlation network, and a new four-factor model is proposed by adding the stock-related structure factor to the Liu et.al. (2019)’s three-factor model. Following the construction of a stock threshold network using the network consistency method, the characteristics of the SRS are extracted based on network analysis. It is found that there is a significant statistical correlation between the characteristics of the SRS and stock return in the Chinese A-share market, and the explanatory power provided by our four-factor model is greater comparing with Fama and French’s three-factor model, verifying the influence of this additional factor on excess stock returns. Our findings not only provide a new perspective in portfolio construction, but also identify a new factor for asset pricing