María Dolores Esteban, Domingo Morales, Agustín Pérez (Operations Research Center, University Miguel Hernández of Elche) and Tomáš Hobza (Czech Technical University in Prague)
Abstract: This paper considers a unit-level multinomial model with logit link for small area estimation of proportion-based socioeconomic indicators. For a labour force survey, some of these indicators are totals of unemployed people, proportions of inactive people, unemployment rates or entropy indexes. A Broyden quasi-Newton algorithm is proposed to calculate the method-of-moment and maximum likelihood estimators of the model parameters. Model-based predictors of small area indicators are derived and their mean squared errors are estimated by parametric bootstrap.