Operational risk is one of the most important risks in financial institutions, however, traditional models couldn' t measure it efficiently as a result of its fat-tail distribution and lacking of data on extreme failure events in practice. The paper introduces a Compound Poisson Process Model to measuring the aggregate extreme operational risk of commercial banks both by General Pareto Distribution ( GPD) and by Generalized extreme value (GEV) distribution. Further, it gets the operational risk data of six international banks using multi-factors Fama-French model to reduce the impact of market risk, credit risk and liquidity risk by considering macroeconomics factors and environments in stock markets. The results show that banks' aggregate extreme operational risk in terms of Conditional Value at Risk ( OpCVaR) are higher than their Value at Risk (OpVaR) in terms of Peak over Threshold (POT), while Block Maxima Models( BMM) can measure the possible maximum value of the loss of banks per day during certain period. Also, we find the aggregate extreme operational risk of Chinese banks are lower than that of international banks if including the systematic crisis but international counterparts have a better risk management than Chinese banks before the crisis.%操作风险管理是当前金融和监管机构密切关注的焦点,但其低频高危数据的严重缺失一直阻碍着极端操作风险度量模型的应用.本文构建了考虑发生频度的复合泊松过程应用模型,从广义帕累托分布(GPD)与广义极值分布(GEV)两种角度度量商业银行的极端操作风险.通过采用国际活跃银行的资本市场数据,基于宏观经济变量影响的Fama-French多因素模型,剥离传统的市场风险、信用风险以及流动性风险等获得综合操作风险极端数据,模拟测度外部事件影响下的银行极端操作风险发现:基于POT(过阈值峰值模型)方法估算的OpCVaR(操作风险的条件在险价值)高于其累积的在险价值,而BMM(分块极值模型)方法提供了一定期限内金融机构的最大损失额频度；且实证分析发现中资银行就其外部宏观经济影响下的极端操作风险而言,普遍低于其国际同行,但排除此次金融危机的系统性影响,则国际活跃银行的风控能力更稳定,这对改善我国银行操作风险管理和监管极具实践意义.
展开▼
