bayesPop: Probabilistic Population Projections
Hana Sevcikova and Adrian E Raftery
October 2013 CSSS Working Paper #135
We describe bayesPop, an R package for producing probabilistic population projections for all countries. This uses probabilistic projections of total fertility and life expectancy generated by Bayesian hierarchical models. It produces a sample from the joint posterior predictive distribution of future age- and sex-specific population counts, fertility rates and mortality rates, as well as future numbers of births and deaths. It provides graphical ways of summarizing this information, including trajectory plots and various kinds of probabilistic population pyramids. An expression language is introduced which allows the user to produce the predictive distribution of a wide variety of derived population quantities, such as the median age or the old age dependency ratio. The package produces aggregated projections for sets of countries, such as UN regions or trading blocs. The methodology has been used by the United Nations to produce their most recent social population projections for all countries, published in the World Population Prospects.
Keywords: Bayesian hierarchical model, Expression language, Population pyramid, United Nations, World Population Prospects