Study On The Application Of Correlated Bayesian Model In Outstanding Claims Reserve Estimation

The assessment of outstanding claims reserve is to estimate the claims amount which is based on statistical data, using mathematical models of insurance companies insured event has occurred but not yet paid or the possible claims amount that future occurrence of the insured to pay. For non-life insurance business, outstanding claims reserve are an important liability on its financial statements. Depositing sufficient reserves for outstanding claims is important to protect the liabilities accruing to the performance of insurance and the interests of the insured. The accuracy of the assessment for outstanding claims reserve will affect the insurance companies operate in a stable and the interests of policy holders, also is high attention by all interested stakeholders and regulatory.For the short time of modern insurance industry and the lack of reliable historical data, the current methods of outstanding claims reserve estimation is usually traditional, such as Chain-Ladder, B-F method, etc. These methods are feasible, but certainty not considered random factors, and estimation accuracy is not high. With the rapid development of China's insurance industry, a variety of new stochastic model is introduced, such as Log-Normal Model, generalized-linear model (GLM), robust Bayesian estimation, etc. These methods improve the precision estimation of outstanding claims reserve, but did not take into account the different development of the correlation between years. However the Correlated Bayesian Model is able to effectively solve this problem, which will make the estimation of outstanding claims reserve more science, more reasonable, and provide a more reliable basis for the risk management and investment strategies of insurance companies.Chapter one is the introduction to explain the significance, domestic and international literature review of thesis research. Simply describes the basic structure of this article analyzes and the innovation of this paper and deficiencies. The innovation is the application in deviation comparison of four models used in this thesis, and the inadequacy of this deviation comparison which can't be a multi-disciplinary comparison is a bit rough. The deviation analysis also needs further improvement.Chapter II analyses the significance, the evaluation to assess the basis and standards, internal and external factors of the outstanding claims reserve. If insurance companies will get more accurate estimation of outstanding claims reserve, we should collect more reliable historical data and practical actuarial valuation methods, study the macroeconomic environment and the company environment such as dynamic changes of various factors. All possible factors should be taking into full account to provide basis for drawing outstanding claims reserve of insurance companies.Chapter III introduces traditional assessment methods that is Chain-ladder model and the B-F model on historical data to estimate outstanding claims reserve, and comments the advantages and disadvantages. Chain-ladder model and the BF model unite in predicting the value of outstanding claims reserve but differ in determining the ultimate cumulative claims. Chain ladder model based entirely on the traffic triangle contained within the paid claims data estimates the ultimate cumulative in each accident claims. And B-F model is completely based on external information to get the ultimate cumulative claims'prior estimates. In other words, in the process of estimating outstanding claims reserve, the standard chain ladder model described above only uses the flow of internal information contained in the triangle, but B-F model also uses part of the external information.Chapter IV is the focus of this paper which studies the application of the relevant Bayesian model in outstanding loss reserve estimation.Using the relevant Gamma distribution, point of view of modern Bayesian analysis applying to the neighboring settlement of the non-independent claims between the amount of years, processes the compensation paid yet, and evaluates the total for all years of outstanding claims reserves. Comparing Bayesian model and traditional model with deviation analysis method obtains the advantages of the relevant Bayesian model in outstanding claims reserve estimation, not only the point estimate, but also under a certain level of confidence interval estimation, with high accuracy relatively, and makes the estimation results more scientific and more reasonable. However, when we use the application of modern Bayesian theory in outstanding claims reserve estimation, we should also consider the shortcomings of model assumptions and deficiencies, taking into account all possible factors that affect the model estimations. Study on the application of correlated Bayesian model in outstanding claims reserve estimation not only provides a reliable basis for the insurance risk management and investment strategy but also makes continuous and stable development in technically supporting the insurance industry, which is the research purpose of this article.Finally, I give the recommendations in application of the model and the drawing of outstanding claims reserve. If insurance companies want to obtain more precise outstanding claims reserve estimation, more reliable historical data should be collected, improving the practicality of the actuarial assessment, and taking into account of the micro factors and dynamic environment factors that may affect the overall results of the outstanding claims reserve assessment. The scientific, effective, accurate and reasonable assessment of outstanding claims reserve are of great significance for the insurance industry's healthy development, economic compensation, and risk management, sustainable development of insurance companies.