Insurance Mode Study Based On Catastrophic Risk Measurement

For a long time, the compensation of catastrophe losses mainly rely on state f financial aid, and the proportion of insurance payments is very low. The Wenchuan earthquake in2008, the compensations of insurance industry insurance are only1.66billion yuan in total, accounting0.196%of all economic losses. This phenomenon caused serious constraints for government, people life and production after the catastrophe. Therefore, in order to reduce the risk of catastrophe losses and improve the catastrophe risk management level, we must summarize the experience of other countries on the theory and practice of catastrophe risk management. We need to establish a sound system of catastrophe insurance through the scientific measure of catastrophe risk and the reasonable pricing.To support the market-oriented operation of catastrophe insurance system, we need to measure the catastrophe risk. The problem of catastrophe risk assessment is actually about what probability distribution the catastrophe loss obey. In this paper, comparative studies were based on parametric methods, semi-parametric method and nonparametric methods. This article chooses earthquake losses between1961and2011as the study sample. We use above data to fit three fat-tailed distributions which are lognormal, weibull, gamma. Then we choose the best fitting distribution by K-S test. Finally, we calculate the premium scales and premium per caput by using VaR. The calculation results can represent the earthquake insurance prices, and can provide the scientific basis for determining the premium rate of earthquake insurance in China. For semi-parametric methods, this article selects the POT model which is one kind of model about extreme value theory. Using this model, we can effective fit the distribution of earthquake disaster loss, and obtain the generalized Pareto loss distribution of earthquake catastrophe.On the basis of the fitting the loss data, then it can calculate the different VaR values under different confidence level, getting a different size and earthquake insurance premium prices, and as a basis for the design of catastrophic insurance risk diversification mechanism. For non-parametric methods, this paper use kernel density estimation to fit the distribution of earthquake catastrophe losses, and we calculate the premium size and per capita premiums under different confidence level on this basis. For these three different fitting methods, K-S test results show that the best fitting result was POT model based on extreme value theory. The final results showed that, although some parameters and some nonparametric methods in a large sample of the loss distribution can give a better fitting, but often give a poor fitting of the tail effect. Therefore, it is difficult to accurately measure of the tail probability of extreme risk, and the POT model in the extreme value theory can use limited extreme data more effectively, and can more accurately describe the distribution characteristics of the tail sequence. What’s more, the form of POT model is simple and easy to calculate, especially in the case of insufficient sample data, which is a more accurate data fitting model.On this basis, by comparison, this paper chooses the optimal distribution of POT model fitting results to calculate VaR risk value, as a basis for the design of catastrophic insurance risk diversification mechanism. And we calculate the insurance premium per people at99%confidence level and it is no more than540yuan, which represent the price of earthquake insurance. This price is acceptable to the residents. Finally, we summarize the main points of all mentioned above, and put forward some specific policy recommendations according to the development of catastrophe insurance in China.