Research On The Reliability Of Reserve Under Different Loss Functions

There is always an insurance company settling compensation for customers at all times in its life.Due to some accidents themselves or the problems of insurance companies,they are not settled in time at the end of the year.In order to solve the above problem of delay in settling claims,insurance companies need to set aside a certain amount of funds in advance to pay future compensation.It is very important to set aside an appropriate number of reserves for the operating conditions of insurance companies.Therefore,it is very important to use suitable models to estimate the more accurate reserve amount.This paper combines the idea of reliability theory under the principle of modified variance premium to solve the problem that classical chain ladder method and Bayesian ladder method rely on known a priori distribution,and it is difficult to find the a priori distribution on which they rely.Through the reliability estimation of the chain ladder factor,the estimated value of the reserve is obtained.Finally,this paper introduces practical cases,and compares with the results of the reserve under the classical chain ladder method,and the results show that the method proposed in this paper is effective.When the traditional B-F model estimates the reserve,the average annual claimed for known accidents is required.Usually,the data come from the a priori estimate of the actuary,which has a certain subjectivity.In order to eliminate this subjective problem,this paper introduces the idea of reliability theory to estimate the average value of accident claims,and obtains the reliability estimate of the annual average accident claim under the B-F model under the secondary loss function,so as to obtain the estimated value of the outstanding compensation reserve.Finally,compared with the traditional B-F method and the chain ladder method,the results show that the method adopted in this paper is feasible.In order to select suitable data,it is sometimes necessary to aggregate individual data to obtain aggregate data in the process of processing data,but some effective information may be omitted in the process of aggregation.In order to solve the problem of leakage of valid data,this paper adopts a bitwise regression model to calculate individual data,and will Abnormal values and the use of insensitive data can be found to obtain more effective prediction results when using a bitwise regression model to directly predict reserves under different components.When the square loss function is used to estimate the reserve,it may occur that more accurate reserve data may not be available due to large claims in previous years,at which time the final desired result can be obtained under the balance loss function by punitive weighting of the data.On the basis of the existing model,based on the parameterized bitwise regression model of GB2 distribution,the estimated value of the reserve is obtained by using the balance loss function combined with the reliability theory.Especially,when = 0,the generalization of existing models is realized.