Probabilistic Population Projection for Countries with Generalized HIV/AIDS Epidemics

Abstract

Population projection has long been an issue for researchers, governments and international organizations so that they can monitor and plan development and resources. The United Nation Population Division (UNPD) publishes the World Population Prospects (WPP) every 2 years, giving estimates of past and projections of future population by age and sex for all countries. The method used for population projection in the 2012 revision of WPP was developed by Raftery et al. (2012), which combines probabilistic projections of life expectancy at birth, fertility and mortality using a cohort component model. It was developed for countries without significant levels of HIV/AIDS prevalence. For countries with generalized HIV/AIDS epidemics, the mortality pattern is different for young and middle-aged adults, and therefore the effect of AIDS should be considered in estimation and projections. In this thesis, we incorporate HIV impact, including HIV prevalence and anti-retroviral therapy (ART) coverage in projecting life expectancy and age- and sex-specific mortality, and use them to probabilistically project population for countries with generalized HIV/AIDS epidemics. Through out-of-sample validation, we demonstrate that our method performs well in population projection for those countries.