Stanford Graduate School of Business
Developed democracies are settling an increased number of refugees, many of whom face challenges integrating into host societies. We develop a flexible data-driven algorithm that optimally assigns refugees across resettlement locations to improve integration outcomes. Using a combination of supervised machine learning and optimal matching, our method discovers and leverages synergies between refugee characteristics and resettlement sites. Backtesting the algorithm on registry data from the United States and Switzerland demonstrates that this approach can lead to considerable gains in refugees’ short- and long-term employment outcomes as compared to current assignment practices. Our data-driven assignment method provides governments with a practical and cost-efficient policy tool that can be immediately implemented within existing institutional structures and offers significant potential to improve refugee integration.