| In recent years, Policy-oriented crop insurance has got more and more attentions by Chinese government and the academic society in China. Academic studies on crop insurance were done more deeply, and the pilot projects of crop insurance have been done in almost every provinces of China. However, setting the premium rate of crop insurance,which is an important point in crop insurance, was not got enough attentions in China till now. The most common approaches to setting the premium rate in China are very rough (assume crop yield fit the normal distribution, or converting the abnormal curve to normal, CATN), and that will lead to great errors in premium rate, which had been proved by the studies and experiences in developed counties.Based on this, in order to make domestic people to be awarded of the importance of insurance actuarial and crop yield distribution, enhance the depth of domestic study on crop insurance, and provide useful suggestions to the development of crop insurance in China, this dissertation do some empirical study on this issue, trying to show the error of premium rate setting will be made by using unfit or'wrong'yield distribution model. Under the support of world famous mathematic and statistical software MATLAB7.0 and EVIEWS5.1, this dissertation makes use of the knowledge of economics, agricultural economics, statistics and econometrics, focus on the studies of three aspects, see follow:a. The principle and approaches of setting premium rate, and the relationship with crop yield distributionAccording to the viewpoint of economics and insurance actuarial principles, I explanted the principle of premium rate setting in theory firstly, and then derived the mathematic relationship between yield distribution and insurance premium from expectation utilityb. The introduction and application of two methods to simulate the yield distribution: parameters and non-parametersAfter review the parametric methods and non-parametric methods presented in prior studies, I chose four parametric models (normal distribution, CATN, Logistic distribution, and Weibull distribution) and Gaussian kernel estimation as the candidate yield distributions to setting the premium rate in my paper. Then, use the maximum likelihood estimation to evaluate the parameters of flexible yield distributions.c. To identify the optimal yield distribution and compare the flexible premium ratesThe criterion to identify the optimal yield distribution is the empirical judgment and Anderson-darling goodness of fit. After AD test to three samples in this paper, it is found that Logistic distribution is the optimal yield distribution for all three samples.In the end of this dissertation, we make the conclusions after empirical research. The first one is that insurance actuarial is necessary to alleviate the adverse selection and moral hazard in the application of crop insurance; the other one is that the accurate of premium rate setting depends on the reasonable and'right'choice of crop yield distributions. And the final one is that the accuracy of CATN is not well, CATN is not a robust method to setting the premium rate, and may not apply to crop insurance. |
PDF Full Text Request