Matching Refugees to Host Country Locations Based on Preferences and Outcomes

Avidit Acharya, Kirk Bansak, and Jens Hainmueller

IPL Working Paper Series, No. 19-03 (2019)


The idea of refugee matching is to select resettlement locations that are likely to be a good fit for a given refugee to thrive. Research has shown that the place of initial settlement has a profound impact on the long-term integration success of refugees [see for example Bansak et al. (2018) included in a previous edition of the literature review]. Two main approaches to refugee matching have emerged: preference-based (i.e. assigned on the basis of the preferences of the refugees or the preferences of the locations), which is uncommon, and outcome-based matching (i.e. maximize refugees’ predicted integration success, e.g. as measured by employment or earnings). The authors summarize the advantages and disadvantages of each method. The authors developed a mechanism that assigns refugees based on optimizing both refugee preferences and expected outcomes. The government first proposes a metric of integration success (e.g. refugee employment, earnings, health outcomes, etc.), and a minimum level of expected integration success that should be achieved, ‘g’. Secondly, refugees are matched to locations based on their preferences subject to meeting the government’s specified threshold. The algorithm then maps preferences to a feasible matching by serially assigning refugees to locations in a way that accommodates their preferences subject to being able to maintain the minimum average level of expected integration success. The authors illustrate this proposed mechanism using simulations and refugee data from the United States. They show that:

  • There is a clear tradeoff between realized preference ranks and outcome scores in all simulations because enforcing the requirement for a higher value of g requires the mechanism to deviate from the preference-based optimization. Conversely, if preferences and outcomes are positively correlated, then matching based on preferences should indirectly also lead to outcome-based matching, and hence deviation from the preference-based solution will not occur until a higher level of g is reached. If, in advance of their preference reporting, refugees were given information on their predicted outcomes in each location, they could incorporate such information into their preference determination, which would help alleviate the tradeoff in the mechanism.
  • The more similar are different families’ preferences, the more rivalrous is the matching procedure, and hence the more difficult it is to match families to one of their top-ranked locations given limited capacity in each location. Given a negative correlation between preferences and outcome scores (i.e. preference-based assignment is counter to the goal of optimizing for realized outcome scores), a positive correlation across families’ preferences has a significant impact on how the tradeoff affects the realized mean outcome score, with the tradeoff being more severe with a low correlation across preference vectors.
  • The more positive is the correlation between preference and outcome vectors, the later the tradeoff kicks in.

The proposed mechanism strikes a compromise by allowing governments to ensure a minimum level of expected integration success while at the same time respecting refugee preferences to the extent possible. It is also strategy proof, does not require refugees to rank all locations, and could be incorporated into existing assignment mechanisms by eliciting refugee preferences for their top locations.