非寿险准备金评估的广义线性模型  

Generalized Linear Models and Their Applications in Non-life Insurance Loss Reserving

作　　者：孟生旺[1] 

机构地区：[1]中国人民大学应用统计科学研究中心,北京100872

出　　处：《统计与信息论坛》2009年第6期3-7,共5页Journal of Statistics and Information

基　　金：教育部新世纪优秀人才支持计划资助

摘　　要：在非寿险准备金评估实务中,保险公司通常应用链梯法和B-F法等确定性模型,但这类模型无法对准备金的预测结果进行统计检验,因此广义线性模型受到了越来越多的关注。在假设增量赔款服从指数分布族的情况下,讨论广义线性模型在准备金评估中的应用,并通过一个实际的流量三角形数据进行实证检验。Non - life insurance oompanies usually apply some deterministic methods, such as chain ladder method and B- F method, to predict their loss reserves. A shortcoming of these methods is that they can not provide a statistical test. Hence generalized linear models have been drawing more and more attention in recent years. The paper assumes that the loss data is from exponential dispersion family, and then a set of practical data is used to construct a generalized linear model and the result shows that the GLM is more flexible and accurate compared with standard chain- ladder method.

关 键 词：非寿险 准备金 广义线性模型 链梯法 指数分布族 

分 类 号：F840.65[经济管理—保险]