Title: Estimating poverty indicators under area-level Poisson mixed models with SAR(1) domain effects
Authors: Miguel Boubeta - Universidade da Coruna (Spain) [presenting]
Maria Jose Lombardia - Universidade da Coruna (Spain)
Domingo Morales - University Miguel Hernandez of Elche (Spain)
Abstract: Poisson mixed models are useful tools for estimating poverty indicators in territorial units, especially when there is a high degree of disaggregation. The number of persons under the poverty line in Galicia (northwest of Spain) is analysed by using area-level Poisson mixed models with SAR(1) domain effects. These models allow a structure of dependence between neighboring domains. In this context, we derive the method of moments (MM) for estimating model parameters and we obtain the empirical best predictors (EBP) of poverty proportions. We compare the EBP against alternative approaches as the synthetic and the plug-in estimators. We use the mean squared error (MSE) as a precision measure of the proposed estimator and we estimate it by parametric bootstrap. Finally, we apply the developed methodology to estimate poverty indicators in Galicia at county level.