GAMLSS模型在车险定价中的应用  

Application of GAMLSS Model in Ratemaking of Auto Insurance

作　　者：刘新红[1,2] 冯媛[2] 米海杰[1,2] 

机构地区：[1]中国人民大学应用统计科学研究中心,北京100872 [2]北京石油化工学院数理系,北京102617

出　　处：《数学的实践与认识》2017年第11期1-8,共8页Mathematics in Practice and Theory

摘　　要：在系统梳理国内外非寿险产品费率厘定方法的基础上,详细介绍了GAMLSS模型,证明了在位置参数和尺度参数的预测中均引入随机效应的GAMLSS模型可更有效地解释纵向数据中个体间的异质性.最后将GAMLSS模型应用于一组纵向车辆保险数据,计算了先验保费、后验保费、后验风险保费和奖惩因子.实证结果表明,GAMLSS模型不仅可为非寿险产品的定价提供依据,而且使风险分类更加稳定、合理.Based on systematic study on the theory about ratemaking of nonlife insuranceploducts in domestic and foreign literature, GAMLSS model is introduced and the hetero-geneity among subjects can be explained more efficiently in longitudinal data. in which theprediction of location and scale parameters has random effect terms is proved. In this paper,GAMLSS model is applied to a group of longitudinal automobile insurance claims data. Thenpriori premium, posteriori premium, bonus-malus factor and posteriori risk premium are com-puted. The result shows that this model is not only rich for ratemaking structure but alsogood for risk classification with stability and rationality.

关 键 词：GAMLSS GB2分布 BCPE分布 

分 类 号：F224[经济管理—国民经济] F840.63