The Combination Of Regression And Time Series Model In The Application Of The Evaluation Of Outstanding Claims Reserve

The outstanding claims reserve is a very important part of the balance sheet if insurance company. Assessing it accurately means a lot to measure company's financial position and assist insurance regulatory commission to supervise insurance companies. The development of insurance in China lags behind its Western rival. Relevant systems and regulations are not perfect enough, the estimate method of claims stay at a preliminary stage, too. More accurate methods are needed as to the quickly development of insurance industry. Until now, many kinds of classical statistical models, such as lognormal model, Bayes method and MCMC method has been applied to the evaluation of outstanding claims reserve. At the same time, time series is applied to the evaluation of outstanding claims reserve too. This paper summarizes above methods, and combine the time series and regression together to estimate the outstanding claims reserve.Firstly, we do a brief introduction to the composition of outstanding claims reserve the deterministic assessment methods, and some influence factor to assessment. Then, we make a detail to the application method of time series of evaluating the outstanding claims reserve. In the empirical part, we use actual indemnity data from 2010Q1 to 2015Q4 of a life insurance company in China. Firstly, we use chain ladder method to estimate indemnity of 2015Q4, and compare it to the actual data of 2015Q4. The result shows that there are large deviation, which will influence the correct decision of the company. Then, a combination of time series and regression is used to predict under three different data b y EVIEWS software. Compare the results of chain ladder model and the decomposed model, we can get the conclusion that the best result is under the condition of the second data scene. Its result has the minimum deviation compared to actual indemnity.