机动车辆保险费率因子测算方法研究-保险学专业论文.docxVIP

I I 摘 要 我国在开放车险市场，在逐渐实施费率市场化的进程中，广义线性模型逐渐 成为非寿险精算定价中的重要统计方法之一，它在车险定价中具有巨大的发展空 间。另外一方面，SAS 作为统计分析功能最强大的商业软件，在国内外保险公司 中广泛使用。 本文通过对大量文献的调研，介绍了非寿险保险产品定价的传统方法以及广 义线性模型在处理保险数据中的实际运用，对财险产品尤其是车险的定价费率因 子的调整和结构进行了详细的介绍与分析。 进一步研究发现，考虑传统广义线性模型存在的缺陷，如保险公司在进行分 类费率厘定时未考虑潜在风险特征的影响，如驾驶人的驾驶习惯等变量。通过引 入个体风险度量方法，利用广义线性混合模型对分类费率厘定技术进行了扩展， 更多得考虑个体保单风险的异质性，通过车险定价中常用的奖惩系统因子进行索 赔次数的风险度量，进一步完善了保险数据定价模型，为今后定价技术发展提供 了参考依据。 另外，本文通过对保险公司（人保财险）和第三方咨询机构（德勤管理咨询） 调研并收集大量保单数据，本文不仅对各定价方法的选择优劣性作出一定的评 价，总结了其中优缺点，还对未来保险公司运用 SAS 软件处理保险数据进一步发 展提出一些建议，同时也为保险公司精算部人员在提供了建模思想的指导。 关键词：产品定价方法；广义线性模型；中国人保财险；SAS 软件 II II Abstract Generalized linear models (GLMs) are the most important tools in non-life insurance pricing, especially for vehicle insurance. In the process of market-oriented reform of insurance rate and open market for vehicle insurance, GLM play the most vital role in the pricing tools. Besides, statistical software, like SAS, has been used for business for a long time, both in finance and insurance area. This paper is based on enormous amount of research and several relative results. We are aimed to introduce some certain numbers of quantitative methods in non-life insurance pricing and how the generalized linear model works in this area. Furthermore, we also conduct the analysis in the structure of rating factors in details. In further study, considering the disadvantage of GLM, for instance, the latent variable can’t be considered in the model, referring to the driving habit in vehicle insurance, to extend the GLM, GLMM introduces random effect to measure the individual risk characteristic. On understanding GLMM technique, we apply BONUS-MALUS system using the technique to claims of vehicle insurance policy combination. This paper as complement to existing pricing method, also remain indicative reference point for the development of risk Measurement. Finally, this paper discussed the present and furfure application of these basic pricing tools, and also describes a real-life case study on using GLM