I. Molina, B. Nandram, J. Rao

A Hierarchical Bayes (HB) procedure is proposed for estimation of general non linear parameters in small areas. An unconventional way of applying HB methods is applied, which does not require the use of Markov Chain Monte Carlo (MCMC) methods. This fastens considerably the HB procedure and avoids convergence problems of the Monte Carlo chains. Only non-informative priors are considered for model parameters. The frequencial validity of this Bayesian method is checked through simulation studies conducted under the frequencial framework, in which the performance of empirical Bayes (EB) and HB methods is compared for specific poverty indicators. The method is applied to the estimation of poverty indicators in Spanish provinces, using data from the Survey on Income and Living Conditions.

Palabras clave: hierarchical Bayes linear mixed models poverty indicators small area estimation

E. González-Dávila, J. A. González-Yanes

A. Satorra Brucart, E. Ventura Colera, C. Rovira, M. García, M. Pardal