Modeling Age-Specific Migration for U.S. States

Crystal Yu, University Of Washington
Hana Sevcikova, University of Washington
Adrian Raftery, University of Washington, Seattle
Sara Curran, University of Washington, Seattle

People often make moves in preparation for, or in response to, major life course events and transitions. This can include entering higher education, obtaining a job, marriage, family, and retirement. However, data on migration events and flows are often scarce and of varying data quality. Moreover, the dynamic nature of migration makes it an extremely difficult phenomenon to empirically study, much less predict. As migration becomes an increasingly important driver of population change in both national and subnational settings, it is imperative to have high-quality, granular migration data, as well as methods for informing and producing migration and population estimates and projections. In this analysis, we draw on previous research on model migration schedules, data from the American Community Survey, and use a Bayesian computational framework to produce probabilistic estimates of age-specific migration rates and counts with prediction intervals for U.S. states. This approach allows us to address potential concerns with migration data availability and reliability, and provides measures of uncertainty surrounding the migration estimates to indicate the uncertainty inherent in predicting migration. These estimates will provide a more complete picture of migration patterns at the state level, and inform probabilistic population projections for all U.S. states.

Keywords: Bayesian methods , Population projections, forecasts, and estimations, Simulation , Internal Migration and Urbanization

See extended abstract.