Estimation of National Labor Indicators Using Additive Log-Ratio Transformed Compositional Fay-Herriot Models

M. J. Lombardía Cortiña, A. Aneiros-Batista, E. López Vizcaíno, S. A. Sperlich

This study introduces area-level compositional mixed models by applying additive log-ratio transformations to the Fay-Herriot framework to estimate labor force indicators at municipal level in Spain. Small area estimators are derived from a bivariate model that accounts for the compositional nature of categorical data, providing estimates for employed, unemployed and inactive people.  The accuracy of estimates is assessed via parametric bootstrap methods, with estimators obtained through restricted maximum likelihood estimation. Labor force survey data are used to illustrate the methodology, demonstrating the model’s ability to produce reliable estimates. This research is part of a broader project funded by the Spanish National Institute of Statistics (INE) under the ETD/503/2021 grant, within Research Line 7.

Keywords: Additive log-ratio transformation Bootstrap resampling Compositional data Fay Herriot model small area estimation Labor force survey.

M. Bugallo Porto, E. Cabello Garcia, M. D. Esteban Lefler, D. Morales, M. Bugallo Porto, S. Rodríguez Ballesteros

M. Bugallo Porto, E. Cabello Garcia, M. D. Esteban Lefler, D. Morales, A. Perez Martín, S. Rodríguez Ballesteros

A. Aneiros-Batista, M. J. Lombardía Cortiña, E. López Vizcaíno, S. A. Sperlich