商车费改背景下汽车保险奖惩系统实证研究

发布时间：2019-06-19 19:01

【摘要】：奖惩系统对于优化交通环境、改善交通秩序以及规范车主的驾车行为、形成良好的驾驶习惯具有积极的促进作用。数据显示,在我国车辆保险的总收入占当年财产险保费总收入的比值远远超过百分之五十,一度达到百分之七十以上。随着我国车辆保有率高速增长,我国发生的交通事故数量以及所造成的财产损失总额一直居高不下。从车辆保险的角度来制定合理的奖惩制度也是改善我国交通秩序的必要的经济手段。同时奖惩制度也在一定程度上对汽车保险的保费收入和赔付支付两方面有着重要影响。近年来我国交通法规对有关的交通违法行为的处罚规则做了重要的修订,同时中国保监会对车险条款和费率管理体制进行了改革。汽车保险费率开始进入深化改革阶段。汽车保险作为重要的财产保险产品之一,费率的厘定基础也是以纯风险保费加上附加保费的基本思路来确定最终保费,奖惩系统根据被保险人以前年度的出险记录,来确定奖惩系统调整系数,在基础保费的计算基础之上对其保费进行一定的调整。本文依据保险公司近年的出险数据,利用贝叶斯方法选择负二项模型作为奖惩系统概率模型,并对此进行了实证分析,建立了考虑索赔次数的保费计算模型,最终得到最优BMS。进而分析了奖惩系统的马尔科夫性,以及达到稳态的过程、稳定保费的情况以及稳态保费。当我们获得稳态保费,也就是相当于在一定程度上可以确定保险公司未来可能收入保费的整体情况,如果我们可以知道保险公司未来可能赔付的情况,就可以对保险公司的财务稳定状况有一个大致的预判和准备。我们利用二项logistic回归的思路,以及保险公司的被保险人数据,选择被保险人驾龄、年龄、性别、年行驶里程数等七个因子,建立出险预测模型,通过该模型预测被保险人的出险可能性,为保险公司预测出险赔付提供思路和方法。在2016年6月,车险改革已经遍及全国各地,本文对我国费改前后实行的奖惩制度,建立相应的马尔科夫转移矩阵,并选取RSAL系数、变异系数、ECL系数等严几个代表性的严厉性指标对系统进行综合性的评价。费改后的奖惩系统更加严厉,BMS级别扩大,奖罚力度增强,从变异系数来看,新的BMS比费改的BMS要严厉许多,但是,新的BMS对于新进入的投保人来说,有着较高的隐形惩罚。最后对于车险费率改革和改善车险奖惩系统提出相关的建议,以促进汽车保险市场的健康发展。汽车保险改革已经推行至全国各地,但是对于改革的说明工作仍然需要重视,汽车保险改革使得风险分级增加,对于优质驾驶员的保费降低,出险次数较多的驾驶员惩罚加大,对于新进入的驾驶员的隐形惩罚也相对提高,这是为了更好的鼓励驾驶员养成良好的驾驶习惯,减少出险事故,而不是为了使出险次数较多的驾驶员退出车险市场,从而造成更大的隐患。除此之外,考虑赔款额在内的模型、更好的利用驾驶员的信息、风险分级的细化,仍然是未来的发展方向,也是本文未来的研究方向。
[Abstract]:The reward and punishment system plays an active role in optimizing the traffic environment, improving the traffic order and regulating the driving behavior of the vehicle owner and forming a good driving habit. The data show that the total income of vehicle insurance in our country accounts for more than 50% of the total income of property insurance premium in the current year, and more than 70%. With the high-speed growth of vehicle tenure, the number of traffic accidents in our country and the total loss of property have been high. The establishment of reasonable reward and punishment system from the angle of vehicle insurance is also the necessary economic means to improve the traffic order in our country. At the same time, the reward and punishment system also has an important influence on the premium income and the payment payment of the automobile insurance to a certain extent. In recent years, China's traffic regulations have made important revisions to the rules on the punishment of traffic offences, and at the same time, the CIRC has reformed the management system of the vehicle insurance and the rate. The rate of car insurance starts to go to the stage of deepening reform. As one of the important property insurance products, the basis of the determination of the rate is to determine the final premium and the reward and punishment system to determine the adjustment coefficient of the reward and punishment system according to the risk record of the insured in the previous year. The premium is adjusted on the basis of the calculation of the basic premium. In this paper, based on the insurance data of the insurance company in recent years, using the Bayesian method to select the negative binomial model as the probability model of the reward and punishment system, the empirical analysis is carried out, and the premium calculation model considering the number of claims is established, and the optimal BMS is finally obtained. Furthermore, the Markov property of the reward and punishment system and the process of steady state, the situation of stable premium and the steady state premium are analyzed. When we get a steady-state premium, that's the case in which the insurance company's future income premium can be determined to some extent, and if we can know what the insurance company might pay for the future, There is a general pre-sentence and preparation for the financial stability of the insurance company. We use two logistic regression thinking, as well as the insured data of the insurance company, to select seven factors such as the driving age, age, sex and annual driving mileage of the insured, to establish the risk prediction model, and to forecast the risk of the insured by the model. And provides a method and a method for predicting the insurance premium for an insurance company. In June,2016, the reform of the vehicle insurance has spread all over the country. This paper establishes a corresponding Markov transition matrix for the system of reward and punishment, which is introduced before and after the reform of the country, and selects RSAL coefficient, coefficient of variation, ECL coefficient and so on to evaluate the system comprehensively. The reward and punishment system after the change is more severe, the BMS level is expanded, and the reward and punishment force is enhanced. In terms of the coefficient of variation, the new BMS is much more severe than the BMS, but the new BMS has a high level of invisibility for the newly-entered policy-holder. Finally, some suggestions are put forward to the reform of the car insurance rate and the improvement of the vehicle insurance and punishment system, so as to promote the healthy development of the automobile insurance market. The reform of the automobile insurance has been carried out to all parts of the country, but the explanation of the reform still needs to be paid attention to, the automobile insurance reform makes the risk classification increase, the insurance premium of the high-quality driver is reduced, the number of the drivers with more risk is increased, The hidden penalty for newly-entering drivers is also relatively improved, which is to better encourage the driver to develop good driving habits, to reduce risk accidents, rather than to exit the risk market for drivers with more risk of risk, resulting in greater hidden dangers. In addition, considering the model of the amount of compensation, better utilization of the driver's information, the refinement of the risk classification is still the future development direction, and the future research direction of this paper.
【学位授予单位】：浙江财经大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F842.634

【相似文献】