In this paper, we introduce a novel centrality measure to evaluate shock propagation on financial networks capturing a notion of contagion and systemic risk contributions. In comparison to many popular centrality metrics (e.g., eigenvector centrality) which provide only a relative centrality between nodes, our proposed measure is in an absolute scale permitting comparisons of contagion risk over time. In addition, we provide a statistical validation method when the network is estimated from data, as is done in practice. This statistical test allows us to reliably assess the computed centrality values. We validate our methodology on simulated data and conduct empirical case studies using financial data. We find that our proposed centrality measure increases significantly during times of financial distress and is able to provide insights in to the (market implied) risk-levels of different firms and sectors.

翻译：本文提出了一种新的中心性度量方法，用于评估金融网络中的冲击传播，以捕捉传染性和系统性风险贡献。与许多流行的中心性指标（如特征向量中心性）仅提供节点间的相对中心性不同，我们提出的度量采用绝对尺度，允许对随时间变化的传染风险进行比较。此外，我们提供了一种统计验证方法，适用于实践中基于数据估计的网络场景。该统计检验使我们能够可靠地评估计算出的中心性值。我们在模拟数据上验证了该方法，并利用金融数据进行了实证案例研究。研究发现，我们所提出的中心性度量在金融困境时期显著增加，并能揭示不同公司和部门（市场隐含的）风险水平。