摘要
本文将HGLMs模型用于未决赔款准备金的评估,应用R软件对未决赔款流量三角形中的增量赔款进行建模,得到未决赔款准备金的估计值和预测误差。这一方法的应用考虑了同一事故年赔款数据反复观测的纵向特征和不同事故年未观测到的特征所导致的异质性,且与贝叶斯方法得到的最佳估计结果非常接近。
The paper uses the hierarchical generalized linear models（HGLMs） for the valuation of outstanding claims reservation, constructing a model of the increased claims in the run--off triangle with R software. This method takes into consideration of the longitudinal characteristics of the data repeated observations in the same accident indemnity and the heterogeneity caused by the characteristics of the different accident years not observed. The result is very similar to the optimal estimation by the Bayes method.
出处
《保险职业学院学报》
2016年第4期83-88,共6页
Journal of Insurance Professional College
