Population projections have until recently usually been done deterministically using the cohort-component method, yielding a single value for each projected future population quantity of interest. Starting in 2015, the United Nations Population Division changed their approach, instead adopted a fully statistical Bayesian probabilistic approach to project fertility, mortality and population for all countries, using methods developed by our group. In 2024, for the first time, uncertainty about net international migration was also included.

In this approach, the total fertility rate, female and male life expectancies at birth, and the net migration rate are projected using Bayesian hierarchical models estimated via Markov chain Monte Carlo. These are then combined with a cohort-component model, yielding probabilistic projections for any future quantity of interest. The methodology is implemented in the bayesPop R package, which has been used by the UN to produce the World Population Prospects since 2015. We have recently extended the method to subnational population projections. I will describe the method and some recent extensions, and illustrate it with subnational demographic data from several countries.