基于时变假设的修正负二项车险索赔频率精算模型

Negative Binomial Claim Frequency Model Modified by Time-Varying Hypothesis in Automobile Insurance

传统车险索赔频率模型都采用风险水平在保险期间保持不变的假设,采用风险水平时变假设,选择Weibull过程作为风险强度函数,引入传统的负二项索赔频率模型。新模型修改原有频域方法为时域参数方法进行参数估计,并使用极大似然估计结合贝叶斯估计的方法估计出Weibull过程的水平参数λ和形状参数β。在β=1时,新模型就等价于传统负二项模型;此外,新模型可为风险上升(β>1)和风险下降(β<1)的保单确定更准确的风险保费。

Traditional claim frequency models of automobile insurance are based on time-constant risk intensity hypothesis.This paper takes time-varying risk intensity hypothesis.Firstly,new model introduces Weibull Process as risk intensity function into traditional Negative Binomial claim frequency model.Then,new model estimates level parameter λ and configuration parameter β of Weibull Process,using time base parameter estimation methods of Bayes Estimation and Most Likely Estimation.If β=1,new model is（equivalent） to traditional Negative Binomial claim frequency model;else,new model can rate more precise risk premium for risk ascending（β〉01） and descending（β〈1） individuals.