New Approach for Simulating Population Outmigration: Combining Linear and Agent-Based Models

Abstract The immediate impact of hazard events is often felt and frequently studied. However, the broader impact of these events in the months and years that follow is an area that requires greater investigation, theorization, and quantification in order to better understand the long-term impacts that such events have on our communities. One topic of great concern is the outmigration of people that is catalyzed by a hazard event such as a flood or tornado. These postevent migration patterns are essential to understanding who is most in need of meaningful assistance following an event to enable them to stay in a community that they have called home. This study focused on modeling these patterns as a combination of a top-down and bottom-up approach to ensure that aggregate community-level data are mirrored while still allowing for emergent behavior to offer greater insight to population flows following a hazard event. In this work, a multiregression model for long-term population migration patterns is put forth and then paired with a household-level agent-based model to capture not only how many people are outmigrating and inmigrating but specifically which people are outmigrating and inmigrating. The new model is illustrated using the community of Mayfield, KY, which was struck by an EF4 tornado. The resultant demographic trends track well with trends shown in the American Community Survey data for Mayfield following the devastating tornado. This model thus shows promise as a path forward for characterizing and ultimately predicting demographic changes following a hazard event. In turn, this information can be used to identify potential policies that will ensure those most at risk of outmigration are allowed the opportunity and provided with the resources they need to remain in their communities following a hazard event.