A0481
Title: Improving small area poverty estimates with random-slope mixed models
Authors: Naomi Diz-Rosales - Universidade da Coruna (Spain) [presenting]
Maria Jose Lombardia - Universidade da Coruna (Spain)
Domingo Morales - University Miguel Hernandez of Elche (Spain)
Abstract: Nowadays, policymakers require detailed socio-economic indicators to assess poverty at very specific levels. However, the challenge arises when sample sizes are small, which affects the precision of estimates. Consequently, a small area estimation methodology is presented to derive predictors of the poverty proportion using random slope mixed models. A Poisson-type area model with random intercept and random slope is introduced, and bootstrap estimators of the mean squared error are defined, both with and without bias correction. A Laplace approximation algorithm is used to calculate maximum likelihood estimators of the model parameters and predictors of random effects. Through simulation experiments, the performance of the fitting algorithm, the predictors and the mean squared error estimators are evaluated. The optimal results obtained allow the model to be applied to real data from the Spanish Living Conditions Survey and the Spanish Labor Force Survey. The final objective is to estimate and map poverty proportions by province and sex in Spain, providing a valuable tool for decision-making in the allocation of resources and policies.
