基于厚尾损失分布的汽车保险定价模型及其应用  

Automobile Insurance Pricing Models Based on Heavy-tailed Loss Distribution and Its Application

作　　者：王选鹤[1] 孟生旺[2] 王雅实[3] 

机构地区：[1]东北财经大学金融学院,辽宁大连116025 [2]中国人民大学应用统计科学研究中心,北京100872 [3]中国政法大学科学技术教学部,北京100027

出　　处：《保险研究》2017年第4期67-78,共12页Insurance Studies

基　　金：国家自然科学基金项目(71601037);国家社科基金重大项目(16ZDA052);教育部人文社会科学重点研究基地重大项目(16JJD910001);中国博士后科学基金特别资助项目(2016T90225);辽宁省社会科学规划基金青年项目(L15CJY006);辽宁经济社会发展课题(2017lslktyb-070)资助

摘　　要：我国目前正在推进商业车险费率的市场化改革,这要求保险公司使用更加准确的风险度量方法和损失预测模型。在汽车保险中,损失的厚尾性对费率厘定和风险管理都具有重要影响。本文引入密度函数的极限方法刻画损失分布的厚尾特征,构建二型广义贝塔分布下的GAMLSS定价模型,以改进传统广义线性模型中的指数族分布假设和只能对均值参数建模的局限。通过对国内商业车险损失数据的实证分析表明,使用厚尾分布假设和GAMLSS定价模型,可以提高汽车保险损失的预测精度,从而厘定更加合理的保险费率。Au quires the use loss is of great tomobile insurance premium rating is currently under a market-oriented reform in China. This re- of more accurate risk measurement methods and loss prediction models for insurers. The heavy-tailed importance for both premium rating and risk management in automobile insurance. This paper intro- duced the limit method of the density function for characterizing the heavy-tailed loss distribution. Thereafter, a GAMLSS pricing model was built based on the generalized beta distribution of the second kind to improve on the ex- ponential family distribution assumption of the traditional generalized linear pricing model and redress the limitation of being able to establish mean parameter modeling only. As indicated by the empirical study based on the domestic automobile insurance loss data, the GAMLSS pricing model, along with the heavy-tailed distribution assumption, can improve the precision for automobile insurance loss prediction, and therefore provide more reasonable premium rating.

关 键 词：汽车保险 费率市场化 厚尾损失 GAMLSS模型 

分 类 号：F340.65[经济管理—产业经济] F224.7