双重广义线性模型在车险费率厘定中的应用研究

发布时间：2018-06-21 09:57

本文选题：车险费率 + 广义线性模型　；参考：《山东科技大学》2017年硕士论文

【摘要】：近年来,随着我国经济发展水平的快速提高以及汽车保有量的不断增加,车险业务的渗透率与覆盖范围进一步扩大,其市场占有率和保费收入在财险业中已跃居首位,是非寿险业务中的第一大险种,车险业务的发展变化对于财险公司而言影响巨大。但是由于国内保险市场建设时间较短,尚不成熟,所以机动车辆保险在发展过程中仍然面临一些困难,而如何对车险进行科学、公平地定价就是其中亟待解决的问题之一。在众多车险费率厘定方法中,广义线性模型(GLM)是现阶段最为常用的一种方法。作为经典线性回归模型的拓展,GLM在理论框架、模型结构以及模型诊断等方面都有了很大的进展,得到了广泛应用。然而在实际建模过程中,GLM依然存在一定的局限性,如关于散度参数的固定常值假设就限制了其在车险定价中的进一步推广。双重广义线性模型(DGLM)作为GLM的扩展模型,为解决上述问题提供了新的思路。DGLM保留了GLM的基本框架,放松了散度参数为固定常值的假设,允许对因变量均值和散度参数同时建立广义线性模型,提高了模型的适应性。本文在已有研究成果的基础上,进一步探讨了DGLM在车险费率厘定中的应用,主要工作如下:(1)对相关理论进行了梳理,分析了将DGLM引入车险定价的合理性。(2)在实证分析中,基于一组真实的车险赔付经验数据,以年均行驶里程数、无索赔优待系统、车型和地区四组变量为费率因子,应用DGLM建立了车险费率的厘定模型,并从参数结果显著性、模型拟合效果以及相对费率三个方面与GLM的费率厘定结果进行了对比分析。结果表明:针对本组数据而言,DGLM得到的费率结构更为合理,符合实际。(3)根据实证结果,参考国内费率厘定工作的现实情况,对于如何提高车险定价的科学性提出了几点建议:①注重数据积累,建立、健全索赔经验数据库;②合理制定费率厘定的原则、目标;③综合运用多种费率厘定方法,创新定价机制。
[Abstract]:In recent years, with the rapid development of China's economy and the continuous increase of vehicle ownership, the penetration and coverage of auto insurance business has further expanded, and its market share and premium income have leapt to the first place in the property insurance industry. The development of auto insurance has a great impact on property insurance companies. However, due to the short construction time and immature of domestic insurance market, motor vehicle insurance still faces some difficulties in the process of development, and how to carry out scientific vehicle insurance, fair pricing is one of the problems to be solved. The generalized linear model (GLM) is the most commonly used method in determining vehicle insurance rates. As an extension of classical linear regression models, GLM has made great progress in theoretical framework, model structure and model diagnosis, and has been widely used. However, in the process of practical modeling, GLM still has some limitations. For example, the constant assumption of divergence parameters limits its further promotion in vehicle insurance pricing. As the extended model of GLM, DGLM provides a new way to solve the above problems. DGLM preserves the basic frame of GLM and relaxes the assumption that the divergence parameter is constant. The generalized linear model is established simultaneously for the dependent variable mean value and divergence parameter, which improves the adaptability of the model. Based on the existing research results, this paper further discusses the application of DGLM in vehicle insurance rate determination. The main work is as follows: 1) combing the relevant theories, and analyzing the rationality of introducing DGLM into auto insurance pricing. Based on a set of real experience data of auto insurance compensation, taking the annual mileage, no-claim preferential treatment system, vehicle type and area as rate factors, a model for determining vehicle insurance rate is established by using DGLM, and the results are significant. The model fitting effect and the relative rate are compared with the GLM rate determination results. The results show that the rate structure obtained by DGLM is more reasonable, which is in line with the actual situation. Several suggestions on how to improve the scientific nature of vehicle insurance pricing are put forward. One should pay attention to the accumulation of data, establish and perfect the database of claim experience. Innovative pricing mechanism.
【学位授予单位】：山东科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F842.634

【参考文献】