Comparison of mean squared error estimatorsunder the Fay-Herriot model: application topoverty and percentage of food expenditure data

Abstract

Small area estimates have received much attention from both private and public sectors dueto the growing demand for effective planning of health services, apportioning of governmentfunds and policy and decision making. The uncertainty of empirical best linear unbiasedpredictor (EBLUP) estimates is widely assessed by mean squared error (MSE). MSEs arecriticized as they are not area specific since they do not depend on the direct estimatorsfrom the survey. In this paper, we compare the performances of different MSE estimatorswith respect to the relative bias and relative risk using a Monte Carlo simulation study.Simulation results suggest the superiority of the proposed MSEs over the existing methodsin some situations. As a case study, the 2010/11 household consumption expendituresurvey (HCES) and the 2007 housing and population census of Ethiopia have been usedto study the performances of the MSE estimators.