基于广义线性模型的车险分类费率厘定研究

发布时间：2018-05-27 14:29

本文选题：分类费率厘定 + 广义线性模型　；参考：《保险研究》2013年09期

【摘要】：在车险分类费率厘定中,广义线性模型已经成为最为常用的方法。广义线性模型着眼于对费率的精确厘定,从而使保险公司的保费收入与其承担的风险相匹配,保障了公司的稳健经营。同时,也为保险公司选择优质业务提供依据,是一项重要的风险管理措施。对广义线性模型在车险分类费率厘定中的应用进行了实证研究,在分析了费率厘定流程的基础上构建了索赔次数和索赔强度的广义线性模型,并对实证结果进行分析。针对模型系数不显著的情况,尝试性地采用了合并风险等级的方法。在实证分析中发现,按照以下原则合并风险等级可以在一定程度上解决风险等级不显著的问题:其一,尽量保证合并后每个风险类别的风险暴露数不至于相差太多;其二,保证每个风险类别的风险暴露数满足信度要求;其三,尽量寻找临近的风险等级合并。
[Abstract]:The generalized linear model has become the most commonly used method in the determination of vehicle insurance classification rate. The generalized linear model focuses on the accurate determination of the rate, so that the premium income of the insurance company is matched with the risk it bears, and the firm's stable operation is ensured. At the same time, it is an important risk management measure to provide basis for insurance companies to select high quality business. This paper makes an empirical study on the application of generalized linear model in vehicle insurance classification rate determination. Based on the analysis of the process of rate determination, a generalized linear model of claim number and claim intensity is constructed, and the empirical results are analyzed. In view of the model coefficient is not significant, we try to adopt the method of combining risk level. In the empirical analysis, it is found that merging risk levels according to the following principles can to some extent solve the problem that risk levels are not significant: first, try to ensure that the number of risk exposures of each risk category after merging is not too different; second, Ensure that the number of risk exposures in each risk category meets the reliability requirements; third, look for adjacent risk levels to merge.
【作者单位】：山东大学经济学院;
【基金】：2013年教育部人文社会科学研究规划基金(项目编号:13YJAZH091)资助
【分类号】：F224;F842.3

【参考文献】