Reconstructing mortality in Brazilian municipalities: methods, limitations, and examples

Bernardo L. Queiroz, Universidade Federal de Minas Gerais (UFMG)
Everton E. C. Lima, Universidade Estadual de Campinas (UNICAMP)
Trifon I. Missov, Max-Planck Institute for Demographic Research
Adam Lenart, Max-Planck Odense Center on the Biodemography of Aging and University of Southern Denmark

The article presents estimated life tables for Brazil at municipal level by applying parametric, non-parametric and correlated frailty models. Direct estimation methods are problematic in small areas due to data quality and the random variations associated with the low number of death counts. The main aim of the article is to produce reliable estimates for life expectancy at birth in small areas as life expectancy is the most widely used indicator to compare mortality among populations and, therefore, of great use for the development and evaluation of public policies. The article shows that, in the case of Brazil, it is essential to combine statistical models with formal demography methods to obtain proper estimates of life expectancy.