基于分位回归的风险保费预测

发布时间：2018-05-02 18:42

  本文选题：保费原理 + 风险保费　； 参考：《统计与信息论坛》2016年09期

【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。
[Abstract]:Risk premium prediction is an important part of non-life insurance rate determination. In the traditional quantile regression determination of risk premium, the quantile level is usually assumed to be given in advance and lacks some objectivity. Therefore, a new method for determining risk premium by quantile regression is proposed. Based on the ruin probability, the total risk premium of the policy portfolio is determined, the quantile regression model of the individual policy is established, and the equality relationship with the total risk premium is established. The quantile level is solved by numerical method, and the prediction of the individual insurance risk premium is realized. A set of actual data analysis shows that the method has a good prediction effect.
【作者单位】： 中国人民大学应用统计科学研究中心;中国人民大学统计学院;
【基金】：国家自然科学基金项目《考虑风险相依的非寿险精算模型研究》(71171193) 教育部重点研究基地重大项目《随机效应模型及其在非寿险风险管理中的应用》(12JJD790025)
【分类号】：F840;O212.1
，

本文编号：1835040