My dissertation enhances and makes use of county-to-county household migration data to investigate several dimensions of interstate and intrastate migration during the 1990 through 2015 period.
The first of three empirical chapters describes an algorithm that combines spatial interaction modelling and linear optimization programming techniques to enhance publicly available county-to-county migration data. The algorithm produces estimates of flows between county-to-county pairs not previously reported due to censoring. The implementation of this algorithm generates an additional approximately 500K records per year. This triples the within-state migration complete case count and increases the interstate migration complete case count by 13-fold. The increased records mean that the migration of an additional approximately 1.6 million households per year, a 30-percent increase, can be incorporated into subsequent analyses. The inclusion of additional records serves to decrease potential biases originating from the omission of data.
The second empirical chapter uses the enhanced county-to-county househol dmigration data to fit production constrained, origin specific, spatial interaction models illustrating the spatiality of households’ destination preferences and trends over time. Households in the Great Plains and the South selected destinations with relatively larger populations while households in western states opted for destinations with relatively smaller populations. In general, households in eastern states moved shorter distances while households in western states moved longer distances. The preference for in-state destinations varies by state: California, Florida and portions of Washington, Texas, and Ohio showed a nearly consistent preference for out-of-state destinations while large swaths of states in the Great Plains and Rocky Mountain regions exhibited preferences for in-state destinations.
The third and final empirical chapter examines the determinants of household movement using a combination of consistent county boundaries, enhanced county-to-county household migration flows, and place-based characteristics such as the age composition of the population, economic, and amenity indicators. County-to-county movement is classified into one of four categories based on the metropolitan statuses of each origin and destination county-to-county pair. The results show how origin and destination place-based characteristics promote outgoing household flows, attract incoming household flows, how these change over time and according to the metropolitan statuses of the origin county and the destination county. Young adults ages 20 through 29 demonstrate the most consistent push and pull factors while other age groups exhibit a variety of patterns over time. Adults ages 60-69 attract households, though this varies by the metropolitan status of the origin and destination counties. Frequently, but not always, an age groups’ ability to promote the outgoing migration of households or attract incoming households reached its maximum or minimum around 2000 with a smaller inflection during the late 2000s. Households are generally attracted to counties with higher annual average pay, more afordable non-metropolitan counties, and non-metropolitan counties with high amenities.