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

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.