基于小波分析的作物区域产量保险费率厘定研究

发布时间：2018-05-07 16:17

本文选题：区域产量保险 + 产量统计模型　；参考：《中国海洋大学》2014年硕士论文

【摘要】：自2007年中央财政开展保费补贴试点以来，我国农业保险在保监会的积极推动下实现了跨越式发展。但是由于信息不对称引发的逆向选择与道德风险问题，造成了传统农业保险的交易成本普遍较高的局面，阻碍了农业保险的发展。20世纪90年代开发的作物区域产量保险（Area-based Crop Yield Insurance）可以有效地避免保险中出现的逆向选择和道德风险问题，在实践中受到各国的青睐。试点区域产量保险是解决我国农业保险保障不足的有效途径，也是未来我国农业保险的发展趋势。然而，科学地厘定保险费率是一个亟待解决的现实问题。因此，本文进行农作物区域产量保险费率厘定的研究对于农业保险及其定价研究的发展具有重大理论意义，同时也为我国农业保险试点提供必要的理论支持。本文首先分析了农作物区域产量保险的特点及我国农业保险试点现状，并立足于我国农作物生产情况，借鉴国外成熟的农作物区域产量保险模式，设计了我国农作物域产量保险体系。其次，通过比较作物区域产量保险费率厘定的相关理论方法的优缺点，选择了产量分布统计模型法作为本文定价方法。在产量分布统计模型法的框架下，针对以往费率厘定方法的缺陷引入小波分析非参数统计方法对定价模型做出了相应的改进：一方面，考虑到小波分析在处理非平稳时间序列分析中体现出的优越性，将其应用于作物产量的趋势分析和预测中。通过小波多辨分析更好地分离周期项、趋势项和随机项，得到反应技术进步的趋势分量和气象波动分量，即将趋势单产（反应技术进步，自然地理环境变化、基础设施改善和劳动者素质提高等原因引起的产量增长趋势）和波动单产（反应气象条件引起的产量波动）相分离。结合支持向量机（SVM）对趋势单产项进行更准确的拟合和预测得到趋势单产预测值。另一方面，在波动单产的分布拟合中，为了解决数据的异方差性，本文采用单产相对损失率这一相对值作为单产波动项的替代值进行概率密度估计。最后，由于非参数方法无需事先假定分布类型，，适用于任意分布的优势，本文采用非参数估计方法对相对单产损失率进行分布拟合得到单产概率密度函数。最终得到改进的基于小波-非参数法的农作物区域产量保险费率厘定模型。选择我国农业大省山东省1949-2011年的玉米历史单产数据为样本进行了实证研究，得到不同保障水平下玉米产量保险费率并进行了比较分析。得到以下结论：通过小波分析法对农作物产量的时间序列进行去趋势处理并分别研究其趋势项和波动项有助于提高费率厘定的准确性；同一保障水平下参数估计正态分布假定下费率水平低于采用非参数核密度估计对单产损失率进行分布拟合得到的纯费率，主要由于农作物单产数据分布往往具有正偏态和尖峰厚尾特征，参数估计正态分布假定下的费率厘定技术低估了风险，非参数法对损失分布的拟合更加符合数据的实际分布也更加准确；随着保障程度的下降其玉米产量保险纯费率呈现下降趋势，合理确定保障水平对作物保险的成功定价和试点发展具有重要意义；另外，本文得出的费率明显低于山东玉米保险实际现行费率。主要由于实际操作中农业保险在查勘定损与理赔等方面的交易成本较高，进而拉高了总保费水平。
[Abstract]:China ' s agricultural insurance has made a leap forward under the active push of the CIRC since the pilot project of premium subsidy in 2007 . However , because of the adverse selection and moral hazard caused by asymmetric information , it has hindered the development of agricultural insurance . The research on regional - based crop yield insurance in the 1990s has great theoretical significance to the development of agricultural insurance and its pricing research , and it also provides the necessary theoretical support for agricultural insurance in China .

This paper first analyzes the characteristics of crop region output insurance and the present situation of agricultural insurance in China , and based on the crop production situation in China , designs the crop field output insurance system in China . Secondly , by comparing the advantages and disadvantages of the relevant theoretical methods determined by crop region yield insurance rates , the paper selects the yield distribution statistical model method as the pricing method in this paper .

On the one hand , considering the superiority of wavelet analysis in dealing with non - stationary time series analysis , it is applied to the trend analysis and prediction of crop yield .

Based on wavelet - non - parametric method , the rate determination model of crop yield in Shandong province was finally obtained . The data of corn yield in Shandong Province from 1949 to 2011 was selected as the sample , and the premium rate of corn yield under different guarantee levels was analyzed . The following conclusions are obtained :

The time series of crop yield was analyzed by wavelet analysis and its trend items and fluctuation items were studied respectively to help to improve the accuracy of rate determination ;
The normal distribution of parameter estimation under the same guarantee level assumes that the lower rate is lower than the pure rate obtained by fitting the non - parametric kernel density estimation on the single yield loss rate .
With the decrease of the level of assurance , the pure rate of corn yield is declining , and the reasonable assurance level is of great significance to the successful pricing and the pilot development of crop insurance .
In addition , the rate of this paper is obviously lower than the actual current rate of Shandong corn insurance .

【学位授予单位】：中国海洋大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F842.66;F326.11

【参考文献】