基于分层广义线性模型的非寿险费率厘定精算模型研究

发布时间：2018-06-18 00:57

本文选题：分层广义线性模型 + 费率厘定　；参考：《统计与信息论坛》2017年06期

【摘要】：非寿险业务中的损失数据结构日益复杂,呈现异质性与相关性并存的异象。分层广义线性模型能够突破传统费率厘定精算方法仅分析风险个体同一保单年损失数据的局限,可以提高复杂结构损失数据预测的准确性。基于分层广义线性模型等方法,研究具有多年损失数据的非寿险费率厘定问题,并以车险和工伤补偿保险的两组损失数据为例进行实证分析。研究结果表明,相对于GLM而言,考虑随机效应后GLMM的拟合优度大幅改善,GLMM与HGLM可以更有效地反映不同风险个体的差异,并有利于揭示风险个体在多个保险期内损失的异质性与相关性。
[Abstract]:Loss data structure in non-life insurance business is becoming more and more complex, showing heterogeneity and correlation. The hierarchical generalized linear model can break through the limitation that the traditional actuarial method can only analyze the annual loss data of the same policy for the individual at risk and can improve the accuracy of the prediction of the loss data of the complex structure. Based on hierarchical generalized linear model, this paper studies the problem of non-life insurance rate determination with multi-year loss data, and takes two groups of loss data of auto insurance and work-related injury insurance as an example for empirical analysis. The results show that compared with GLM, GLMM and HGLM can reflect the differences of different risk individuals more effectively after considering random effect. And it is helpful to reveal the heterogeneity and correlation of risk individuals' losses in multiple insurance periods.
【作者单位】：天津理工大学管理学院;南开大学金融学院;中南财经政法大学金融学院;
【基金】：国家自然科学基金青年项目《驾驶行为对车险市场索赔的影响、风险评估与激励策略研究》(71603180);《基于不完全信息和相依结构的最优再保险策略研究》(71601186);《基于相依结构的多元索赔准备金评估随机性方法研究》(71401041)
【分类号】：F224;F840.4

【相似文献】