Using synthetic farm data to estimate individual nitrate leaching levels
提出一种生成合成农场数据的方法，结合贝叶斯网络和非参数回归，估算希腊色萨利地区农场的个体硝酸盐淋溶率，并设计最优税收方案以减少化肥污染。
Abstract This article delineates a synthetic population generation scheme in an attempt to estimate individual nitrate leaching rates among Greek farms in the region of Thessaly. The proposed scheme relies upon the construction of a Bayesian network describing farming activities in the region, which, coupled with the use of nonparametric regression models, facilitate the consistent generation of synthetic farm data. Then, building upon the sequential generalized maximum entropy approach suggested by Kaplan et al., enhanced with the multiple production relations model proposed by Murty et al., we obtain econometric estimates of the unified farm production and nitrate leaching technology for the synthetic population of farms. The estimation of individual nitrate emissions leads, thus, to the formulation of an optimal taxation scheme aiming to mitigate the negative externality created by chemical fertilization in agricultural activities.