Classification Ratemaking Of Automobile Based On Generalized Linear Models

As the development of China's economy and the improvement of people's living standards, motor vehicle insurance business has been growing rapidly. Since 1988, motor vehicle insurance, whose premiums were more than the premiums of the enterprise property insurance for the first time, has been the first of the property insurance. In 2010, motor vehicle insurance premiums accounted for 75.30% of the property insurance income. As an important pillar of China's insurance industry, the motor vehicle insurance has a great impact on the future of the property insurance. Moreover,the operating results of the motor vehicle insurance will directly affect the profitability of insurance companies, as well as the sustainable development of China's insurance industry.The scientific ratemaking is the key of the profitability and development of the motor vehicle insurance. With the development of the property insurance business and the accelerating process of opening up, market-oriented reform of auto insurance rates will be the inevitable choice and trend of China. The achievement of the self-pricing of the motor vehicle insurance largely depends on accurate and scientific ratemaking. In China, the non-life actuarial technology is underdeveloped. The generalized linear model which is widely used in ratemaking of motor vehicle insurance attracted the attention because of its disadvantages; even it has the virtue of the fitting linear with insurance data. Based on the facts discussed above, the choice of a new model which is used as the analysis method, so as to improve the ratemaking's accuracy, has become a realistic problem that should be resolved properly in the development of China's motor vehicle insurance.This paper used finance, actuarial science and statistics as the theoretical basis of theory,and systematically analyzed the ratemaking of automobiles with the ways of the combination of qualitative research and quantitative research, theoretical research and empirical research. First, this article describes the background of the problem which has an important impact to China's insurance industry. The motor vehicle insurance is an important part of the insurance industry and the ratemaking is the key of auto insurance market-oriented reform. Therefore, the problem of ratemaking of auto insurance is of great significance for the insurance industry, the financial industry and even the stable development of the economy. Secondly, the paper introduces the fundamentals of ratemaking. This part describes the concept and the importance of the ratemaking, as well as the factors and constitution of the premium rate. In addition, the part describes the traditional ratemaking methods and their defects. In the third pat, the paper compares the traditional linear model, generalized linear model and generalized linear mixed model. The comparison includes advantages and disadvantages, the conditions of uses of models and the statistical diagnosis of models. The result is that generalized linear mixed model has obvious advantages. In the empirical part, this paper selects generalized linear mixed model as analysis method, through the empirical analysis of a group of auto insurance data, verified the accuracy of generalized linear mixed models in ratemaking.All in all, motor vehicle insurance is an important pillar of China's insurance industry. Given the importance of independent pricing of auto insurance products, China should gradually adopt more precise methods for classification ratemaking. So the CIRC should strengthen supervision and establish a good competition mechanism. And the association of insurance industry should establish an industry database as soon as possible in order to improve the accuracy of the actuarial. The insurance companies should also improve training. Only in this way, China's motor vehicle insurance and China's insurance industry can develop stably and orderly.