RBNS的线性预测模型  

Linear Prediction Model for RBNS

作　　者：俞雪梨[1,2] 吴贤毅[1] 

机构地区：[1]华东师范大学统计与精算学系,上海200241 [2]上海金融学院公共经济管理学院,上海201209

出　　处：《应用概率统计》2011年第2期194-209,共16页Chinese Journal of Applied Probability and Statistics

基　　金：上海市哲学社会科学基金(2010BJB004);国家自然科学基金(71071056)资助

摘　　要：传统的准备金方法都是基于聚合数据的,聚合数据是个体数据的汇总,它们丢失了许多有用信息,影响了准备金预测的准确性.本文提出了一个基于个体数据的线性预测模型,该模型不需要对数据的矩的具体形式进行假设,更不需要对数据的分布进行假设,而只需假设个体索赔数据的前两阶矩存在,具有适用范围广,简单易操作等特点.在文章的最后,通过随机模拟把提出的方法与著名的链梯法进行了对比,模拟结果显示,本文提出的方法是行之有效的.The traditional claims reserving approaches are all based on the aggregated data and usually produce inaccurate predictions of the reserve because the aggregated data usually failed to use all the information of the individual claims. A linear prediction model based on the individual claims is proposed in this papery in which only the second order moment of the data is required. This method is flexible and simple and hence easy to apply in reality. The proposed method is compared with the traditional chain ladder method via a simulation study as well, and the simulation results illustrate that the proposed approach performs well.

关 键 词：准备金 RBNS 线性预测 链梯法 

分 类 号：O213.9[理学—概率论与数理统计]