中美上市银行高维网络结构特征与系统性风险传染

Characteristics of High-dimensional Network Structures and Systemic Risk Infection in Chinese and US-Listed Banks

基于2008年1月1日至2017年12月31日30家中美上市银行的数据,运用Graphical LASSO算法估计和可视化了中美银行间的高维网络结构关联度,同时运用网络中心性、簇系数、小世界效应指数、网络密度等指标刻画银行网络拓扑结构特征和银行间系统性风险传染路径。研究发现:总体上中美银行之间的关联度较弱,系统性风险跨境传染的路径较少;大型银行及系统重要性银行在网络中的影响力较强,部分股份制银行也具有较强的信息溢出效应和调节能力,并且具有一定的小世界效应;金融危机或股灾期间,跨国银行网络的小世界效应和网络密度会增强,国内银行网络的小世界效应也会略有提升,系统性风险的传染路径也随之拓宽。强有力的救助计划或调控政策可以有效抑制,甚至削弱银行网络中的小世界效应,防止系统性风险通过关键节点向其他银行蔓延。

Based on the data of 30 listed banks in China and the US from January 1,2008,to December 31,2017,this study applied the graphical lasso to estimate and visualize the correlation of high-dimensional network structures between Chinese and US banks. It used indicators such as network centrality,cluster coefficients,small-world effect,and network density to illustrate the topological characteristics of banking networks and inter-bank systemic risk infection paths. In general,the results showed a weak correlation between Chinese and US banks and limited paths of cross-border systemic risk infection. Large banks and systemically prominent banks have stronger influence on networks. Some joint-stock banks also display strong information spillover effects and moderation ability,indicating a small-world effect. During financial crises or stock market crashes,the small-world effect and network density of multinational banking networks are strengthened. The small-world effect of the Chinese banking network is also slightly improved,while the paths for systemic risk infection are broadened. A strong rescue plan or regulation policy can effectively suppress or even weaken the small-world effect in banking networks and prevent systemic risks from spreading to other banks via key nodes.