Bayesian Estimation and Mortality Forecast in Small Areas of Brazil between 2010 and 2030

Marcos Gonzaga, Universidade Federal do Rio Grande do Norte (UFRN)
Bernardo L. Queiroz, Universidade Federal de Minas Gerais
Everton E. C. Lima, IFCH - UNICAMP
Flávio Henrique M. de A. Freire, Universidade Federal do Rio Grande do Norte
José Henrique Monteiro da Silva , Universidade Estadual de Campinas (UNICAMP)

Reliable estimates of mortality rates by sex and age are important for planning, evaluating and allocating public health resources. This demand is become even bigger at more disaggregated geographic levels. However, the accuracy of estimates of mortality rates by sex and age in sub-national populations, especially in developing countries, is still a challenge for demographers and population and health scholars, especially in the context of incomplete vital records and coverage differentiated by age. To deal with this limitation, we propose new methodological alternative to obtain estimates and forecast mortality in scenario of defective data. The procedure involves incorporating uncertainties in the estimates and projections of specific mortality rates at sub-national populations. The uncertainties are related to the two main sources of error: low number of people exposed to death and incomplete coverage in vital records. We use a Bayesian model to produce probabilistic estimates of mortality rates by sex and simple ages for the municipalities of Brazil in 2010. Afterwards, we apply the original model proposed by Lee and Carter (1992) and an improvement of the model is presented to project these rates, making it possible to estimate municipal mortality tables in Brazil for the period 2010-2030. Our results show that the methodological options used were effective in achieving the proposed objectives.

See paper

 Presented in Session P3. Poster Session Migration, Economics, Environment, Methods, History and Policy