The increasing frequency and severity of natural disasters across the United States present significant risks and threaten the long-term sustainability of living conditions. Migration, a key adaptive response to these threats, varies significantly based on individual characteristics, particularly educational attainment. This study employs a structural model that differentiates migration patterns by skill level, revealing that low-skilled individuals are more likely to relocate following natural disasters. A distinguishing feature of this model is its consideration of how individuals weigh future disaster risks when choosing potential destinations. The findings indicate that conventional disaster damage estimates, which are often influenced by socioeconomic factors, tend to underestimate migration for both high- and low-skilled groups. By accounting for the bilateral nature of migration and isolating damage estimates from external influences, the study shows that an MSA experiencing a one percentage point increase in disaster damage is likely to see a corresponding 1 percentage point decrease in its high-skilled population over a five-year period.