| The outstanding claims reserve of non-life insurance business is an important basis of funds reservation and capital finance account, also the biggest liabilities item in the Balanced Sheet of non-life insurance companies. So an exact assessment of the outstanding claims reserve has an important meaning to the efficient control of insurance regulators and the risk prevention, capital finance account of insurance company. As our insurance industry started relatively late, the legal system of insurance-related is not perfect, the method of extracting outstanding claims reserve is also backward, and we also use some traditional method, with consideration being not comprehensive enough. With the development of our insurance industry and gradually expanding, it's necessary to bring in some method that can accurately assess the reserve.In the past twenty years, the assessment of outstanding claims reserve is an important research field of insurance. Many statistical models like Linear regression model, Loglinear model, Mack model, Generalized linear models, Credibility Theory model, Bayesian Model and Bootstrap method are applied to the assessment of outstanding claims reserve. At the same time Time Series technology has matured and now it is possible to apply Time Series to estimate the outstanding claims reserves. This paper summarizes the virtue of recent research, and brings in the Regression and Time Series combined model, then analyzes the assessment of outstanding claims reserves using the data of a domestic insurance company.Chapter one is an introduction of all the paper that describes the background, significance of the topic, also summarizes domestic and international research on the status of outstanding claims reserve. On this basis, this section gives the basic framework, the innovation and deficiencies of this paper.Chapter two summarizes the element, the impact factors and the significance of the non-life insurance business reserve, also expounds the Chain-ladder Method, B-F method of traditional assessment. But for some new business with insufficient data and the some insurance business which is vulnerable to abnormal compensate, the Chain-ladder Method is not reliable. Yet the B-F method joined the parameter values which is based on external factors and experience of actuary avoids the influence in a certain extent which comes from less data or larger error.Chapter three has three parts. Section one detailed the Generalized linear models and the application in the outstanding claims reserve. Here this paper also gives a method how to estimate the parameters and prediction error, which is precisely the shortcomings because the prediction error is approximate and it is complex to calculate. Section two detailed the Bootstrap method and its application in the outstanding claims reserve. This method solves some question commendably on how to solve prediction accuracy. Section three introduces the Time Series model and summarizes the nature and method in order to identify of ARMA. Then the linear regression model and Time Series model are combined to apply into the assessment of outstanding claims reserve. Fitting the residual part of the linear regression model with ARMA can improve the effectiveness of regression parameters and predict reserve better.Chapter four predicts a group of real domestic claims data separately by Bootstrap GLM and Regression-Time Series combined model,then the model's applicability is illustrated. Finally the paper discusses the next further consideration and some proposal on the assessment of outstanding claims reserve for our country. Based on previous research and through demonstration and analysis this paper concludes that when there are fewer cases in the sample data Bootstrap GLM can previous the outstand claims reserve better with the valuation of stability, free from man-made factors, and the advantages of simple operation by computer, but if there are enough samples Regression-Time Series combined model has a better prediction. |
PDF Full Text Request