The Feasibility of Using Data-Driven Algorithmic Recommendations for Refugee Placement in Norway

Abstract

A growing body of research suggests that refugees’ initial settlement area can have a long-run impact on subsequent integration outcomes. As a result, matching refugees and asylum seekers to initial locations where they are likely to succeed holds the potential to improve their labor market integration. In this report we focus on the GeoMatch algorithm, which is a recommendation tool that provides settlement officers with data-driven location recommendations for incoming refugees and asylum seekers. Leveraging machine learning on historical data, the tool predicts labor market outcomes for individuals across possible settlement areas. A flexible allocation algorithm then provides location recommendations for each family unit while taking capacity constraints into account. Drawing on administrative data from Statistics Norway and incorporating a set of realistic constraints, we find that using GeoMatch recommendations could improve refugees’ monthly earnings by up to 55% over baseline. The report ends with a discussion of how the tool can be implemented in the Norwegian context.

The Feasibility of Using Data-Driven Algorithmic Recommendations for Refugee Placement in Norway