A Fay-Herriot Model for Estimating the Proportion of Households in Poverty in Brazilian Municipalities
Viviane Quintaes1, Nícia Hansen1, Denise Silva1, Djalma Pessoa1, Pedro Silva2
1Coordination of Methods and Quality - COMEQ, Brazilian Institute of Geography and Statistics - IBGE, Rio de Janeiro, Brazil; 2National School of Statistical Sciences - ENCE, Brazilian Institute of Geography and Statistics - IBGE, Rio de Janeiro, Brazil

The Brazilian Institute of Geography and Statistics (IBGE) faces the challenge of producing comprehensive, accurate and reliable information under financial and time constraints. The pressure for reducing sample sizes and respondent burden reveals increases the need for methods to produce small area statistics from combined data sources. Small area estimation covers a variety of methods used to produce survey-based estimates for geographical areas or domains of study in which the sample sizes are too small to provide reliable direct estimates. This work presents a first attempt to implement an area level Fay-Herriot model to estimate the proportion of households in poverty for municipalities of Minas Gerais state combining survey data from the Family Expenditure Survey with Census and administrative data.

This work is part of a project aiming at developing small area estimation models and procedures to allow publication of statistical outputs for areas or domains that are not currently considered for publication due to the low precision of the estimates (or lack of sample observations in some of these areas or domains).

IBGE has already published a first set of poverty estimates based on the so-called World Bank Method (Elbers, Lanjouw, and Lanjouw, 2002). The focus of this work is to test some well-known small area estimation models that have been used successfully by other National Statistical Institutes and compare results with those obtained under the World Bank Method.

The paper describes the steps taken for developing this area level model. Results obtained so far proved the feasibility of the area level approach and the vital role of good quality auxiliary data but also indicated the need to revise the choice of target areas for estimation in order to avoid the use of synthetic estimator to obtain estimates for large numbers of areas with no sample observations.

Keywords: Fay-Herriot model; small area estimation; poverty mapping; Brazilian Family Expenditure Survey

Biography: Nícia Hansen graduated in 2007 from the Federal University of Rio de Janeiro with a degree in Statistics and holds a MSc in Statistics from the same University. She works as a methodologist in the Brazilian Institute of Geography and Statistics (IBGE) in the Coordination of Methods and Quality (COMEQ) and has experience in Bayesian inference, spatial statistics and small area estimation.