Network Connectedness And China'S Systemic Financial Risk Contagion--An Analysis Based On Big Data

Excessive network connectedness among financial institutions will amplify financial shocks through contagion effects. In order to measure the correlation between institutions from multiple channels more comprehensively, this paper adopts empirical Bayesian method to integrate the stock return correlation, sentiment correlation and marginal expected shortfall correlation of institutions, which respectively reflects the risk changes of the financial system, the internet public sentiment and the degree of risk contribution. More than 3.72 million posts are crawled from the Eastmoney Net through the Web Crawler Technology to obtain investors' sentiment. The Graphical Gaussian model is used to measure the partial correlation coefficients between every two institutions and the heat map is used to visualize them. The force-directed graph is used to construct the static and dynamic network of financial institutions. We also analyze the systemic risk contagion annual network and how the contagion propagates within sectors and across sectors. On this basis, the systemic importance of financial institutions is analyzed. The results show that China's systemic risk contagion has cluster effect; The network correlation of financial institutions has certain stability and time-variability; The systemic risk contagion level of different sectors has different characteristics and trend; The correlation and size are important factors affecting the systemic importance of financial institutions. This research has important reference value for the regulatory agencies to carry out the retrospective macro-prudential and targeted supervision.