Pull your small area estimates up by the bootstraps

This paper presents a methodological update to the World Bank's toolkit for small area estimation. The reviews computational procedures of current methods used by institution: traditional ELL approach and Empirical Best (EB) addition introduced imitate original EB procedure Molina Rao [Small estimation poverty indicators. Canadian J Stat. 2010;38(3):369–385], including heteroskedasticity survey weights, but using different bootstrap approach, here referred as clustered bootstrap. Simulation experiments provide empirical evidence shortcomings which yields biased noisier point estimates. document an Bank’s implementation considering noise estimation, extended complex designs heteroscedasticity. illustrate that revised yield considerably less more efficient estimators than those obtained from approach.