A1429
Title: Small area estimation of employment indicators under area-level mixed models
Authors: Domingo Morales - University Miguel Hernandez of Elche (Spain) [presenting]
E Cabello - Universidad Miguel Hernandez de Elche (Spain)
MD Esteban - University-Miguel Hernandez of Elche (Spain)
Tomas Hobza - Czech Technical University in Prague (Czech Republic)
Agustin Perez - University Miguel Hernandez of Elche (Spain)
Abstract: Three area-level model-based procedures for predicting unemployment rates are analyzed. The first procedure fits a multinomial model to direct estimators of the total numbers of employed and unemployed individuals in small areas. The second procedure adopts a compositional data analysis framework and fits a multivariate normal model to log-ratio transformations of direct estimators of domain-specific category proportions. The third procedure is a recent approach based on the Dirichlet distribution and constitutes the main focus of the analysis. After model selection and fitting, predictions of small-area proportions, totals, and unemployment rates are obtained, and their mean squared errors are estimated using a parametric bootstrap approach. A series of simulation studies is conducted to assess the performance of the fitting algorithm, the small-area predictors, and the bootstrap estimation method.The methodology is illustrated using data from the Spanish Labour Force Survey for the fourth quarter of 2022, with the objective of estimating provincial proportions of employed, unemployed, and inactive individuals, as well as unemployment rates, disaggregated by sex and age group. Comparisons among the three approaches are also presented.
