分野別セミナー

The area-level model has played a vital role in both the theoretical development and practical implementation of small area estimation. In this study, we propose a multiple-goal empirical best linear unbiased predictor (EBLUP) under an area-level model that simultaneously: (i) improves the estimation of shrinkage factors, (ii) addresses the issue of over-shrinkage in EBLUP, and (iii) eliminates the need for complex bias correction procedures when constructing strictly positive, second-order unbiased mean squared error (MSE) estimators, whether using Taylor series expansion or a single parametric bootstrap method. The proposed methodology is evaluated through a Monte Carlo simulation study and applied to real data. This is a joint work with Prof. Partha Lahiri of the University of Maryland, College Park.