A Study On Credibility Method Of Non-life Actuarial Science Based On Markov Chain Monte Carlo Technique

With the five year period after China's entry into WTO coming to an end, Chinese financial industries have fulfilled their commitment of opening to outside world on the whole. Insurance, which is the completely open one in financial industries, has to face the fierce competition. Therefore, how to make reasonable premiums which can both satisfy regulatory requirement of China Insurance Regulatory Commission and maintain competitive in price has become the highlight of actuarial science. In this paper, we have a study on credibility method which is an extensively used experience rating method in today's property and casualty insurance.There is a close relationship between exact credibility and Bayesian method. Exact credibility premium is a particular case which confines Bayesian estimation of risk premium in linear combination of claim experiences. Since MCMC technique is the mainstream method in Bayesian statistics, we tentatively combine exact credibility model with MCMC technique in order to solve the deficiency of previous model which can only give point estimation of credibility factor.In this paper, the significance of rate making technique is introduced from the viewpoint of fierce competition in insurance industry. Meanwhile we express definition and principle of experience rating method. Then, we give a thorough examination at credibility method, including aspects of history background, conception, application and calculating method. After analysis of Bayesian method in exact credibility theory, it is time to put forward the new approach. On understanding MCMC technique, we apply exact credibility model using the technique to claims of vehicle insurance policy combination. Comparing with existing methods, the new approach which secures the reasonable outcome can also give interval estimation of credibility premium and credibility factor, which provides basis of experience rate making for actuary. Finally, a summary is made on result and weakness of this paper as well as further research in credibility theory.