基于非参数核密度估计方法的环境责任保险保费厘定研究

发布时间：2018-04-15 07:39

本文选题：环境责任保险 + 保费　；参考：《中国海洋大学》2013年硕士论文

【摘要】：近年来,环境污染问题在经济高速增长的背景下日益突显,大气污染、海洋污染、淡水污染等突发事件频繁发生。对环境污染的受害人进行赔偿,是保护公民合法环境权益的必然要求,但是巨额的损害赔偿金又给企业和社会带来了沉重负担。为解决这一困惑局面,环境责任保险制度应运而生。2007年12月,原国家环保总局和中国保监会联合印发了《关于环境污染责任保险工作的指导意见》,2013年1月中国环境保护部和保监会又联合发布《关于开展环境污染强制责任保险试点工作的指导意见》,为环境污染责任保险制度的发展提供了政策依据。保险定价是环境责任保险制度的重要内容之一,保险定价主要包括三个方面内容：保费的厘定,准备金的提取以及再保险安排,其中保费厘定是保险定价的核心内容。保费的高低直接影响企业的参保意愿与保险公司的经营状况,因此,科学地厘定保费不仅有利于环境责任保险制度的有效运行,同时有利于企业和社会的稳定发展。我国环境责任保险制度起步较晚,对于环境责任保险保费厘定方面的研究较少。目前学者对环境责任保险保费厘定方法的研究中,大都采用参数估计方法对索赔金额数据的分布进行拟合,即预先假定索赔金额数据符合某一理论分布,然后进行保费和费率的厘定。而非参数估计方法的思想为在对总体分布尽可能少的假定前提下,利用数据本身对总体分布进行估计。在研究大量金融领域与非寿险领域其他保险业务运用非参数方法进行定价的文献的基础上,考虑到环境责任保险索赔金额数据的分布大多具有尖峰、厚尾、不完全对称的特征,不能完全符合某一理论模型,本文将非参数估计理论中的非参数核密度估计方法运用到环境责任保险保费厘定过程的研究中。主要的研究工作有： (1)在对环境责任保险的概念、属性和保费厘定特点进行分析的基础上,对环境责任保险的保费厘定过程进行概括,可分为索赔数据的整理及分布拟合、纯保费的计算及费用和利润附加三个方面,重点分析了索赔金额数据分布的拟合这一步骤。论文分析了基于参数估计方法的环境责任保险索赔金额数据分布的拟合方法,并将非参数估计方法引入到环境责任保险保费厘定的过程中,最后对两种估计方法进行了比较。 (2)对非参数估计理论中的非参数密度估计理论进行了系统的分析,通过对比分析非参数密度估计理论的四种估计方法—直方图估计法、Rosenblatt估计方法、Parzen核密度估计方法和最邻近估计法,发现Parzen核密度估计法应用范围较广,理论较为完善,应用性较强,因此选择此方法进行实证分析。在研究了非参数核密度估计理论中窗宽的选取原则及两种常用的选取方法——Silverman经验法则和交叉验证法的基础上,结合环境责任保险数据的有限性及方法的适用性,选取Silverman经验法则进行实证研究。 (3)以化学原料与化学制品制造业为研究对象,对环境责任保险的保费厘定进行实证研究。首先对索赔次数的分布进行拟合。然后对索赔金额的概率函数进行估计：分析环境责任保险的索赔金额数据的分布特征,由于其不符合Silverman经验法则的适用条件,需要对数据进行变换处理；对处理后的数据使用Silverman经验法则选取窗宽,应用非参数核密度估计方法对其概率密度函数进行估计,并且对估计的结果进行检验；运用数学变换得到索赔金额数据的概率密度函数。最后根据索赔金额的密度函数得到索赔金额的期望,并结合索赔频率计算得出保费。
[Abstract]:In recent years, the problem of environmental pollution in the background of rapid economic growth has become increasingly prominent, air pollution, marine pollution, water pollution incidents have occurred frequently. To compensate the victims of environmental pollution, is an inevitable requirement to protect the lawful rights and interests of the environment, but also huge damages to the enterprise and society has brought a heavy burden. In order to solve the confusion situation, environmental liability insurance system came into being.2007 in December, the former State Environmental Protection Administration and China CIRC jointly issued "guidance on the work of environmental pollution liability insurance on January 2013, China Ministry of environmental protection and the China Insurance Regulatory Commission jointly issued the" guiding opinions on carrying out the "environmental pollution liability insurance pilot work that provides the policy basis for the development of environmental pollution liability insurance system.
Insurance pricing is one of the important contents of the environmental liability insurance system, insurance pricing mainly includes three aspects: the determination of the premium, reserving and reinsurance arrangements, which is the core content of insurance premium pricing. The premium will directly affect the willingness and insurance companies operating conditions, therefore, scientifically determining the premium not only conducive to the effective operation of the environmental liability insurance system, and is conducive to the stable development of enterprises and the society. The environmental liability insurance system in China started late, for the study of environmental liability insurance premium of less. At present, researches on the environmental liability insurance premium method, distribution fitting method for estimating the amount of the claim data most of the parameters, which presupposes the claim amount distribution data in accordance with a theory, then the premium and the premium rate. . the non parameter estimation method for as much as possible to the overall distribution of the premise of less, using data on the overall distribution is estimated. Based pricing literature in a number of studies of the financial sector and the non life insurance business in other fields by using non parametric methods, considering the distribution of environmental liability insurance claims data mostly with spike thick tail characteristics, is not completely symmetrical, not fully consistent with a theoretical model, this paper will study the nonparametric estimation of nonparametric kernel density estimation method is applied to the theory of environmental liability insurance premium is determined in the process. The main research work includes:
(1) the concept of environmental liability insurance, analyses the attributes and characteristics of premium, the premium of environmental liability insurance are summarized, which can be divided into fitting and finishing distribution of claim data, calculate the net premium and additional expenses and profits of the three aspects, analyzes the steps the claim amount distribution fitting. This paper analyzes the fitting parameter estimation method of environmental liability insurance claims data distribution method based on the non parameter estimation method is introduced into the process of environmental liability insurance premium in the two estimation methods are compared.
(2) the theory of systematic analysis in the theory of nonparametric density estimation of non parametric estimation, through the comparative analysis of non parametric density estimation theory of four estimation methods: histogram estimation method, Rosenblatt estimation method, Parzen kernel density estimation method and the nearest neighbor estimation method, Parzen kernel density estimation method has a wide application range and the relatively perfect theory and practical application, so we choose this method to do empirical analysis. On the basis of the principle of non parametric kernel density estimation and two kinds of selection methods -- Silverman experience and cross validation method to select the window width in the theory, combined with the limited applicability and method of environmental liability insurance data the selection of Silverman rule of thumb for empirical research.
(3) the chemical raw materials and chemical products manufacturing industry as the research object, carries on the empirical research on the environmental liability insurance premium. The distribution fitting of the claim number. Then the probability function of the amount of the claim to be estimated: the analysis on distribution characteristics of the number of claims of environmental liability insurance according to the applicable conditions, because it does not meet the the Silverman rule of thumb, the need for data transform; selecting the bandwidth using Silverman rule of thumb after dealing with the data, methods of the probability density function of non parametric kernel density estimation and application, and to verify the estimation results; using mathematical transform probability density function of the claim amount of data. Finally, the amount of the claim expectations are according to the density function of the amount of the claim, and combined with the claim frequency calculated premium.

【学位授予单位】：中国海洋大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F842.69

【参考文献】