In this paper,Edgeworth expansion for the nearest neighbor-kernel estimate and random weighting approximation of conditional density are given and the consistency and convergence rate are proved.
基于Neyman-Rubin潜在结果框架,构建k近邻核估计量来测度响应变量随机缺失情形下的条件平均处理效应(conditional average treatment effect,CATE),旨在评估不同处理方式对个体的影响.证明了k近邻核估计量的几乎完全收敛性和渐近正态性...基于Neyman-Rubin潜在结果框架,构建k近邻核估计量来测度响应变量随机缺失情形下的条件平均处理效应(conditional average treatment effect,CATE),旨在评估不同处理方式对个体的影响.证明了k近邻核估计量的几乎完全收敛性和渐近正态性.数值模拟表明k近邻核估计量的表现优良,利用真实数据进行实证分析,实证结果显示k近邻核估计量具有较小的平均绝对偏差和均方根误差.展开更多
文摘In this paper,Edgeworth expansion for the nearest neighbor-kernel estimate and random weighting approximation of conditional density are given and the consistency and convergence rate are proved.
文摘基于Neyman-Rubin潜在结果框架,构建k近邻核估计量来测度响应变量随机缺失情形下的条件平均处理效应(conditional average treatment effect,CATE),旨在评估不同处理方式对个体的影响.证明了k近邻核估计量的几乎完全收敛性和渐近正态性.数值模拟表明k近邻核估计量的表现优良,利用真实数据进行实证分析,实证结果显示k近邻核估计量具有较小的平均绝对偏差和均方根误差.
