To better adjust pro-agricultural policies and the relevant fiscal subsidy policies, it is important to measure the impact of policy agricultural insurance on the production behavior of peasant households. Measuring moral hazard can be really tough since it is hard to peel heterogeneity of a sample. By natural experiment, this study separate moral hazard and adverse selection of insurance. The results find that the existence of agriculture insurance notably increase the death rate of hogs. Besides, having hog insurance will increase peasants’ report willingness of severe epidemic situation, while decrease their efforts on risk management. This study provide experimental basis to better adjust policy agricultural insurance. Besides, the study peels heterogeneity of the sample, making its research results more accurate, which may theoretically expand the current researches.