A Data-Driven Approach to Refugee Integration

More than 270 million people today have left their home countries to seek a better life abroad. However, once they land in a new country, immigrants may face unemployment or end up in jobs that are a poor fit for their skills and work experience. These economic hurdles in turn can make it more difficult for them to integrate into their host community: for example, to learn the local language, develop social connections, and feel a sense of belonging. Importantly, there is existing evidence demonstrating the importance of a refugee’s initial location for their integration outcomes (e.g. Åslund and Rooth 2007; Damm 2014; Mossad et al. 2018; Marten et al. 2019).  This research suggests location-based strategies and policies can pave the way to greater integration and social inclusion of immigrants. Furthermore, these synergies that exist between refugees and locations can be uncovered using supervised machine learning (Bansak et al. 2018).

The Immigration Policy Lab is partnering with the Central Agency for the Reception of Asylum Seekers in the Netherlands (COA) to implement a data-driven matching algorithm tool called GeoMatch based on the insight that immigrants are more likely to successfully integrate into their new communities if their initial location is a good fit. The algorithm helps placement officers identify the best locations for incoming refugees that optimize on their integration outcomes. The team conducted backtests to demonstrate the potential for improved outcomes, such as higher earnings or employment rates, and the next step is to launch pilot programs to test whether these potential gains can be realized in practice. This research partnership provides an exciting opportunity to test this algorithm through a randomized controlled trial (RCT); implement this tool as part of its refugee reception program if the evaluation is successful; and generate rigorous evidence on the impact of this matching algorithm to scale to additional countries.