大宗农产品价格风险评估——基于小波神经网络-Bootstrap方法的实证研究

Evaluation on Price Risk of Bulk Agricultural Product: Empirical Study Based on Wavelet Neural Network-Bootstrap Method

构建了一个将小波神经网络与Bootstrap抽样相结合的价格风险评估模型。采用国际通用的VaR(在险价值)风险指标评估了国内小麦、水稻、玉米、大豆和棉花5种主要大宗农产品现货价格的风险水平,仿真研究了以上大宗农产品价格下跌风险和价格上涨风险的分布特征。结果表明:按价格风险水平由高到低对5种主要大宗农产品进行排序依次为棉花、大豆、玉米、小麦和水稻;从风险均值来看,我国大宗农产品价格特别是粮食价格的风险处于较低水平;从风险的经验分布来看,除大豆外,其他大宗农产品(特别是小麦、水稻和玉米)的涨价风险高于跌价风险;5种农产品的价格均存在偏度风险和峰度风险。

This paper combines wavelet neural network with bootstrap method to construct a risk evaluation model. And then it uses the international indicator,VaR（value at risk）, to evaluate the risk levels of spot prices of domestic five kinds of bulk agricultural products including wheat, rice, corn, soya and cotton, and analyzes the distribution characteristics of left and right tails risk. The results show as follows：it is sorted by VaR in descending order that is cotton,soya,corn,wheat and rice;according to the mean value of VaR,the price risk level of domestic bulk agricultural products is low,especially grain crops;according to the distribution characteristic of VaR, the value of right tail risk is bigger than that of left tail risk,except soya;there exist skewness risk and kurtosis risk in the prices of domestic five kinds of bulk agricultural products.