School of Mathematical Sciences, Queen Mary University of London, London, E14NS, United Kingdom.
Dipartimento di Fisica e Astronomia, Università di Catania and INFN, I-95123, Catania, Italy.
Sci Rep. 2018 Apr 3;8(1):5561. doi: 10.1038/s41598-018-23689-5.
The interconnectedness of financial institutions affects instability and credit crises. To quantify systemic risk we introduce here the PD model, a dynamic model that combines credit risk techniques with a contagion mechanism on the network of exposures among banks. A potential loss distribution is obtained through a multi-period Monte Carlo simulation that considers the probability of default (PD) of the banks and their tendency of defaulting in the same time interval. A contagion process increases the PD of banks exposed toward distressed counterparties. The systemic risk is measured by statistics of the loss distribution, while the contribution of each node is quantified by the new measures PDRank and PDImpact. We illustrate how the model works on the network of the European Global Systemically Important Banks. For a certain range of the banks' capital and of their assets volatility, our results reveal the emergence of a strong contagion regime where lower default correlation between banks corresponds to higher losses. This is the opposite of the diversification benefits postulated by standard credit risk models used by banks and regulators who could therefore underestimate the capital needed to overcome a period of crisis, thereby contributing to the financial system instability.
金融机构的相互关联性会影响不稳定性和信用危机。为了量化系统性风险，我们在这里引入 PD 模型，这是一个将信用风险技术与银行间风险敞口网络的传染机制相结合的动态模型。通过多期蒙特卡罗模拟获得潜在损失分布，该模拟考虑了银行的违约概率（PD）及其在同一时间间隔内违约的趋势。传染过程会增加面临困境的交易对手的银行的 PD。通过损失分布的统计数据来衡量系统性风险，而每个节点的贡献则通过新的 PDRank 和 PDImpact 措施来量化。我们展示了该模型如何在欧洲全球系统重要性银行网络上运作。对于银行资本和资产波动性的特定范围，我们的结果揭示了一种强烈的传染机制的出现，即银行之间较低的违约相关性对应着更高的损失。这与银行和监管机构使用的标准信用风险模型所假设的多样化收益相反，他们因此可能低估了克服危机时期所需的资本，从而导致金融体系不稳定。
