Research Of Risk Interval Methodology For Non-life Insurance Outstanding Reserve Estimation

In 2007, the spreading subprime mortgage crisis swept through the Wall Street and the world financial institutions. The turbulence of AIG maked us view the risk management of insurance industry much more rational. As a crucial role in the financial system, insurance company will play an important role in the future.Non-life insurance is insurance business except life insurance, included by property insurance, liability insurance, credit guarantee insurance, short-term health insurance, accident injury insurance and reinsurance of the business above. The reserve of the non-life insurance is the capital that company should reserve or prepare for future according to insurance contract. The reserve of the non-life insurance is always the important problem in the actuarial science. The research of the past tends to find the better ways to catch the evaluation of reserve, which got the models developing from certainty to randomness. Based on the chain-ladder method, the introducing of modern statistical model such as state space model, generalized linear model, Bayesian theory, MCMC simulation method, Kalman filtering diversified reserve estimation.This paper analyzed the management of reserve from the perspective of risk early warning, which discussed the risk interval of reserve. Compared to the traditional thinking, the risk interval plans to control and manage:firstly, this paper would catch the fluctuating range of historical data and build the interval; secondly used the so-called risk interval to predict the trend of claims, in order to make rational reserve decision. This paper gave the definition of risk interval, and discussed the assumptions, implementation plan and risk parameter, and then it investigated the estimation error as supplement. This paper did empirical research using the risk interval and chain-ladder method. It drew the conclusion that the concept of risk interval method would be more rigorous and more applicable to solve the condition that month claims changed frequently. This method gives a criterion for managers, especially on the condition that the changes in recent months separate from the historical risk interval, which helps actuarial managers to prevent the risk before it really arrives.