基于分位回归的风险保费预测 |
| |
引用本文: | 杨亮,孟生旺.基于分位回归的风险保费预测[J].统计与信息论坛,2016(9):83-88. |
| |
作者姓名: | 杨亮 孟生旺 |
| |
作者单位: | 中国人民大学 应用统计科学研究中心,北京100872; 中国人民大学 统计学院,北京100872 |
| |
基金项目: | 国家自然科学基金,教育部重点研究基地重大项目 |
| |
摘 要: | 风险保费预测是非寿险费率厘定的重要组成部分。在传统的分位回归厘定风险保费中,通常假设分位数水平是事先给定的,缺乏一定的客观性。为此,提出了一种应用分位回归厘定风险保费的新方法。基于破产概率确定保单组合的总风险保费,建立个体保单的分位回归模型,并与总风险保费建立等式关系,通过数值方法求解出分位数水平,实现对个体保单风险保费的预测。通过一组实际数据分析表明,该方法具有良好的预测效果。
|
关 键 词: | 保费原理 风险保费 分位回归 Tweedie回归 |
Prediction of Risk Premium Based on Quantile Regression |
| |
Abstract: | Prediction of risk premiums is an important part in the non-life insurance premium ratemaking.During the determining the risk premium in the traditional quantile regression,quantile level is generally given in advance,which is lack of obj ectivity.Therefore,we propose a new method for determining the risk premium by applying of quantile regression.Firstly,determining the total risk premiums based on the probability of bankruptcy;secondly,to establish quantile regression model in individual policies,and establish relationships with total risk premium;finally,to solve for quantile level by numerical methods,obtain the individual risk premiums.Basing on a set of actual data,the results demonstrates that the method has high forecasting accuracy. |
| |
Keywords: | premium principle risk premium quantile regression Tweedie regression |
本文献已被 万方数据 等数据库收录! |
|