A machine learning-based algorithm can substantially improve employment prospects for refugees over current approaches, easing their transition as they become accustomed to a new home. After experiencing war and years of displacement, refugees arrive in a new country with few resources and must adjust to an unfamiliar society. Among the many challenges they face, refugees often experience low levels of employment in the years following their arrival. In the hopes of boosting employment success, Kirk Bansak et al. developed an algorithm that uses a combination of supervised machine learning and optimal matching to align refugees with the best employment locations, given their individual skillsets. For example, French-speaking refugees benefit more when relocated to French-speaking regions, compared to German-speaking regions. The algorithm is trained using historical data that captures refugees' background characteristics (e.g., country of origin, language skills, gender, age, etc.), time of arrival, assigned location, and measured employment success. To test the performance of their algorithm, the researchers used recent data from historical registries from two countries with different refugee assignment regimes and refugee populations - the U.S. and Switzerland. In the U.S., they found that applying the algorithm was likely to boost the prospect of employment by 41%, compared to current assignment practices, and in Switzerland, it boosted prospective employment by 71%, over the status quo. These gains, say the authors, come at little cost to host country governments. The results provide governments with a practical tool for improving refugee integration over status quo practices.