Research On Evaluation Method Of Non-life Insurance Claim Reserve Based On Dependent Structure

The loss reserving of a non-life insurer is a reasonable estimate of the insurer's future indemnity.The reasonableness of the estimate is not only related to the solvency and stability of the insurance company,but also affects the short-term and long-term of the insurance company.In general,the methods of estimating loss reserving of non-life insurance are generally deterministic methods,which can only obtain the point estimate,and can not take the correlation between actual data into account.For this reason,this paper will expand to the randomness method based on the deterministic method.Considering the dependence between different types of data in the single line of business and the dependence between different lines of business,which can increase the adaptability and predictive ability of the model.At the same time we can simulate the distribution of the characteristics of the reserve.In this article,we firstly review the traditional methods for evaluating the outstanding claims reserve,we summarize the existing main evaluation methods and show stochastic expectation payment method based on traditional expectation payment method.The method of nuclear density is used to realize the randomization of the development factors in the expectation payment method.We can obtain continuous data based on the deterministic point data,and the smooth curve is used to fit the feature distribution of the reserve to increase the flexibility of the model.In the kernel density estimation,different reserve models can be obtained through different choices of kernel function and combined bandwidth,and no specific hypothesis distribution is needed.Secondly,on the basis of the stochastic expectation payment method,considering the dependence between the cumulative paid claims and the cumulative reported claims,in the process of simulating the outstanding claims reserve,the Copula function is combined with the expectation payment method.The Copula function is used to reflect the correlation between paid claims and reported claims data.At the same time,we can get the predicted distribution of the outstanding claims reserve,the mean,standard deviation and quantile.Thirdly,aiming at the reserve evaluation of the portfolio in practice,a reserve evaluation model based on Sarmanov dependent structure is proposed.Considering the random effects and simulating the interdependence between the indemnity flow triangles,the dependence based on different lines of business is reflected in the reserve.In the assessment,the model can better estimate the reserve.Finally,conducting empirical analysis of the proposed models,through the specific data in the actuarial practice,using R language and other tools to get the corresponding results,and give an analysis of the results,which can verify the validity of the models.