Estimating employment-related migration from overlapping migration and commuting networks

A U.S. county workplace-to-workplace or latent migration data set is generated from overlapping migration and commuting networks. The latent migration network is the estimated number of movers between places of work, which is then compared with the actual number of migrants between places of residence. This allows both employment-related and amenity-related migration and pull/push factors as causes of migration flows to be identified and contrasted. Certain counties and cities that are not important migration destinations (e.g., with <200,000 net in-migrants between 1995 and 2000) according to official data are in fact important targets when additional in-migrants who commute into surrounding counties also are considered. An econometric analysis is then used to examine whether different regressors have different effects on the residence- versus employment-based migration patterns. This is a first assessment of whether or not the proposed approach has merit. Results are consistent with prior expectations regarding the factors that would motivate latent versus actual migration.