关键词: 复杂网络, 网络结构, 风险敞口, 皮尔逊相关系数, 格兰杰因果检验

Abstract:

In recent years， financial markets have become extremely volatile， especially the global financial crisis in 2008 and the continued global plunge in global stock markets caused by the COVID-19 in 2020. This has drawn lots of attention from academia trying to measure systemic risks and grasp the system risk spread across sectors or markets. The complex relationship between financial markets and their internal elements is the carrier of systemic risk transmission， and their connectedness patterns or structures play an important role in the formation and infection process of systemic risks. For the interconnectedness within a market， once one sector encounters a risk shock， the risk will affect other sectors through strong linkages and contagion mechanisms， and even spread to the entire financial markets. China is currently in a critical period of supply-side reform and economic transformation. As international financial markets become increasingly connected， domestic financial market reforms are gradually deepening and financial innovations are changing rapidly. As the second largest market in the world， Chinese financial system is increasingly attracting attention from countries around the world following a series of liberalisation policies. Also note the unevenness of the development of China's financial sector and the differences in its contribution. In this context， investigation into the connectedness among financial markets and the systemic risk spillovers contagion mechanism across sectors or markets become important and necessary.Pearson correlation coefficients and Granger causality tests are used to construct undirected and directed industry networks， CoVaR models are used to calculate industry risk， and quantile regressions are combined with to explore the interrelationship between industry risk and network structure.With the help of module analysis， the clustering of industry networks under different extreme events and the transmission paths of risks between modules are analysed in depth， while the effect of network structure on industry exposures is examined. The results show that：Both industry networks show a tendency for finance to become the centre of the network， and the directed network constructed by Granger causality test can better explain industry risk. Financial market shocks can have an impact on the structure of industry networks， with different mechanisms for the spread of industry networks under different extreme events. There is also a significant impact of network structure on industry risk， with reductions in industry network clustering coefficients and global efficiency and increases in meso-centrality reducing industry exposure.The findings of this paper have a certain value of participation in financial risk prevention and industrial structure enhancement.

Key words: complex network, network structure, riskexposure, Pearson correlation coefficients, Granger causality test