Accounting for migrant definitions improves global migration modeling

Reliable estimates of international migrant populations matter for migration policy, social cohesion, and integration. Most countries define migrants as foreign-born population, but about one in five defines them by citizenship. Under the citizenship definition, migrants are foreign residents who exit the migrant stock by acquiring the host country's citizenship. Existing migration models apply the country-of-birth definition to all countries due to data availability and reduced modeling complexity. We present a stochastic model that consistently applies the migrant definition of each country and represents citizenship acquisition as a probabilistic process based on legal residence requirements. Across 192 countries from 1990 to 2020, we find that accounting for migrant definitions improves the agreement of simulation results with UN reported migrant stocks. The normalized error against reported stocks decreases from 35.1% to 23.0% in Africa and from 18.1% to 13.6% in Asia, the two world regions where the citizenship-based migrant definition is most common. In lower-middle- and upper-middle-income destinations, the 95% prediction intervals cover the reported stocks in all reference years. Results in regions that apply a country-of-birth migrant definition remain essentially unchanged. Where reported figures are available, estimated citizenship acquisitions are plausibly represented within our model. Our results show that migrant definitions, and the citizenship laws behind them, shape the dynamics of migrant populations and not only how these populations are counted. The approach is transferable to other global migration models and can be refined as citizenship acquisition data become available for non-OECD countries.