Premium subsidy system is an important part of policy agricultural insurance. China has set up policy agricultural insurance system for several years. The object is "low premium & wide coverage and low protection & guaranteed costs". But it cannot be realized if the premium subsidy system is unreasonable. Based on the theory of income and consumption in economics, an income-consumption model of policy agricultural insurance is built in the article according to the characters of Chinese rural market. Now, the premium subsidy system of policy agricultural insurance in China adopts fixed premium subsidy ratio with threshold. Peasants must pay a part of the premium otherwise they cannot get the premium subsidies from governments, which is called "no participation, no subsidies". In such system, the income-consumption model shows that the consumption of policy agricultural insurance will increase firstly and decrease later if the peasant family income increases, while the consumption of commercial insurance will increase with the increase in the peasant family income. The real premium ratio of commercial insurance keeps dropping and is obviously lower than the actual premium rate of policy agricultural insurance, as those commercial insurance companies always provide different kinds of discounts or convenient services. So, two hypothesizes are made according to the results of the theory model.

Using the data from "Thousands Villages Investigation" (TVI) of Shanghai University of Finance and Economics (SUFE) in 2015, we test these two hypothesizes of the theory model. In every summer holiday, SUFE organizes teachers and students to join TVI, which began from 2008. By 2015 it has lasted eight years. The investigation topic of 2015 is "the coverage and usage of basic finance services in rural market". 30 teachers and 1 418 students investigated 31 provinces in China except Hongkong, Taiwan and Macao. Finally, there are 663 valid village questionnaires and 14 339 peasant household valid questionnaires. Some statistical models are used to analyze the data of TVI. Some adjustments are made in order to change those qualitative data from questionnaires to quantitative data for statistical models. LOWESS is used firstly to intuitively show the relationship among consumption of policy agricultural insurance, consumption of commercial insurance and peasant family income. OLS regression model is used later to get the deep relationship among those variables. With the data analysis, we find that the relationship is really looked as a reversed U between the consumption of policy agricultural insurance and peasant family income. However, the relationship is monotonically increasing between the consumption of commercial insurance and peasant family income. So, not only theory analysis, but also empirical analysis of TVI test these two hypothesizes. The object of policy agricultural insurance in China cannot be made true if we continue to carry out the policy of fixed premium subsidy ratio with threshold. The consumption of policy agricultural insurance would decrease and even eventually disappear with the increase in peasant family income. Therefore, it is very important to reform the premium subsidy system in China.

Excess layered subsidy system is a very good suggestion to reform the premium subsidy system in China. It is a little like regressive ratio system called by other scholars. The premium subsidy ratio would decrease with the increase in the protection of policy agricultural insurance. For example, the increase in the protection of policy agricultural insurance from 40% to 100% leads to the decrease in the premium subsidy ratio from 100% to 20%. In this way, those poor families will get basic protection with low premium. They need not to pay any more by themselves. But those rich families would pay more if they want to get higher or full protection. The income-consumption model theoretically verifies the effectiveness of excess layered subsidy system. Then, the detailed system is designed according to the real data of China rural market. Both show that excess layeredsubsidy system can improve the coverage scale and overall protection of policy agricultural insurance effectively.