Optimizing placement to improve refugees’ outcomes

Newly arrived refugees often struggle for many years before they are able to find employment in the host country. This can be explained both by individual background characteristics, as well as the location where they live. Importantly, some areas might be a better fit for certain individuals. In Sweden, refugees who are unable to find housing on their own are assigned to municipalities more or less randomly.

This project examines if it is possible to improve the matching between refugees and municipalities in order to increase refugees’ employment prospects. Using machine learning we will estimate each family’s probability of employment in each municipality, in order to predict where they will be most successful. We will also compare how assigning individuals to achieve the highest employment rate affects other outcomes, such as language training and study participation.