摘要
内部欺诈已成为中国银行业操作风险最重要的风险源,假设损失属性会对损失强度产生影响,在此前提下分析可能影响内部欺诈损失强度的损失属性(机构属性、人员属性和事件属性等)。假设模型为广义线性模型,采用交叉验证法判别影响因素。损失属性对内部欺诈损失强度确有影响,各属性对损失强度的影响各不相同:事件属性中的内部员工参与人数属性对损失强度影响最大,其次是人员属性的职位属性、性别属性和事件属性的次数属性,而其他属性对损失强度影响不大。
Internal fraud has been the most important factor leading to operational risk in Chinese banks. This article believes that the loss severity is subject to the loss attributes and so we analyze some potential loss attributes （institute attributes, staff attributes and event attributes）. Suppose that the model is Generalized Linear Model, and by using cross validation method, we find out that some loss attributes indeed influence the loss severity, and the influences are different. The number of staff involved has the biggest influence on the loss severity, followed by the attributes of position, gender and the number of frauds. The influence of other attributes is not that big. If the fraud happens many times, and a number of staff is involved, including executives, and the dominant players are males, it usually leads to serious loss. Thus, the operation with many staff involved, including executives should be frequently monitored.
出处
《山东财政学院学报》
2011年第4期23-27,共5页
Journal of Shandong Finance Institute
基金
山东省自然科学基金高校
科研单位联合项目(ZR2010GL011)
关键词
操作风险
内部欺诈
广义线性模型
交叉验证
operational risk
internal fraud
generalized linear model
cross validation
