It is shown that the weighted residual-based estimator of Schick, Zhu, and Du (2017) is efficient in some special cases and can be made to be efficient by adding a stochastic correction term. The efficiency is shown by deriving the efficient influence function and establishing a uniform stochastic expansion with this influence function. The correction term relies on estimators of the score function for the errors and other characteristics of the model.

中文翻译：

---

可变系数回归模型中误差分布的有效估计

结果表明，Schick，Zhu和Du（2017）的基于加权残差的估计量在某些特殊情况下是有效的，并且可以通过添加随机校正项使其变得有效。通过推导有效影响函数并用该影响函数建立均匀的随机扩展来显示效率。对于模型的误差和其他特征，校正项依赖于评分函数的估计器。