Penalized Weighted Least Squares to Small Area Estimation
提出一种惩罚加权最小二乘法用于单元水平模型的小区域估计，统一了传统经验最优线性无偏预测与考虑抽样权重的伪经验最优线性无偏预测，并对模型误设具有稳健性。
Abstract In this paper, a penalized weighted least squares approach is proposed for small area estimation under the unit level model. The new method not only unifies the traditional empirical best linear unbiased prediction that does not take sampling design into account and the pseudo‐empirical best linear unbiased prediction that incorporates sampling weights but also has the desirable robustness property to model misspecification compared with existing methods. The empirical small area estimator is given, and the corresponding second‐order approximation to mean squared error estimator is derived. Numerical comparisons based on synthetic and real data sets show superior performance of the proposed method to currently available estimators in the literature.