This paper explores alternative approaches to the projection of international migration. Instead of relying on the residual concept of net migration or a migrant pool model of migration flows, an approach is suggested that uses estimates of international migration flows in an extended multiregional projection model. The contributions to the study of international migration are threefold. (1) The paper suggests that international migration must be modeled and projected as the interaction between sending and receiving entities (world regions), an aspect often present in subnational and internal migration studies, but virtually absent in the field of international migration projections. (2) The paper shows that the classic formulation of the demographic multiregional model is biased toward emigration, as it omits the other side of the migration process. Therefore, the multiregional model needs to be amended to include immigration (or admission) as well. Inspired by well-established approaches in nuptiality analysis that employ mating rules or marriage functions, the paper suggests using a migration transfer function that formally captures the interaction between sending and receiving entities, countries, or regions. Simple transfer functions are suggested. (3) The paper demonstrates the feasibility of an extended multiregional model with a range of migration scenarios (emigration- dominant scenario, immigration-dominant scenario, harmonic mean interaction scenario balances potential emigration and potential immigration/admission. A comparison suggests that the scenario that implements interaction is more realistic. An annex presents the mathematics of interacting migration flows and the reformulation of how to impose age structure on incomplete migration data.
See methodology brief here.