The Choise Of Agricultural Reinsurance Pattern

China's politicail agricultural insurance was developed rapidly since 2007,but in 2008,some natural disaster such as the snow strom and flood in southern China bringed a heavy loss to the agricultural insurance industry. The systematic feature of natural risk makes the large number theory invalid in the agricultural insurance industry. An agricultural insurer which sells its insurance goods in a particular area may suffer a lot if it can't diversify its business risk efficiently. Based on this reason, the article analyeses the effect of agricultural reinsurance in diversifing the risk of the insurer. We provide an agricultural reinsurance pattern which is suitable for China based on the foreign experience of developing an agricultural reinsurance system. Using the data collected from the tobacco insurance in Xingshan county,Hubei province, the author fits distributions on loss ratio data of tobacco insurance. Based on the result of fitting and taking insurance company, agricultural reinsurance organization and government as a whole, the tobacco reinsurance pattern of Xingshan County is applied and the effect of agricultural reinsurance is proved.Conclusions of this article are:Firstly, considering the conditions of China and the specificities of various agricultural reinsurance provision pattern, This article argues that China Reinsurance(Group) Corporation can provide and manage agricultural reinsurance efficiently thus it can be the main supplier of agricultural reinsurance in China, also the government can act as the final reinsurer and make contributions to the development of agricultural reinsurance in China.Secondly, The average loss ratio of the tobacco insurance in Xingshan county is 104.65%, the insurance company still provide the service though slight defilit happens. The fitting result shows that the distribution of simple loss ratio follows a Weibull distribution whose shape parameter is 1.4959 and the scale parameter is 1.0909. The distribution of combined ratio follows a Gamma distribution whose shape parameter is 2.305 and the scale parameter is 0.4717.Both of the distributions have a long tail show a probability that a fierce loss ratio occurs.Thirdly, Based on the specificities of tobacco insurance in Xingshan County, the author argues that the stop-loss reinsurance is the reinsurance pattern which is suitable for Xingshan County. In the tobacco reinsurance pattern in Xingshan county, the reinsurance company will provide a two-layer stop-loss reinsurance to the combained ratio over 110%. The government will cover all the risk once the loss ratio is over 260%. After buying agricultural reinsurance, the expected loss ratio falls to 0.8169 from 1.0872, agricultural reinsurance works well in diversifing the risk.