(CN) – Quotas and security concerns often color debates in nations affected by the global refugee crisis. What happens once refugees arrive, however, is rarely considered, and finding jobs and integrating are typically difficult due to language barriers and discrimination in their host nations.
To limit the extent of these challenges, a team of researchers has developed a new algorithm that can help governments and resettlement agencies place refugees in cities or towns that are well-suited for them based on the refugees’ characteristics and data on thousands of others who came to the host nation before them.
The tool, which is described in a report published Thursday in the journal Science, reveals clear connections between refugees’ traits and local conditions. Some displaced individuals’ strengths will be rewarded more in certain places, while characteristics that might be issues in some locations become less of a problem in others.
Currently, these factors are not considered in any systematic way. Refugees who come to the United States tend to be placed where there’s space to house them at that specific moment. In other nations like Switzerland, asylum seekers are dispersed randomly and proportionally.
While both nations have data on how earlier arrivals fared economically, the power of this information to aid future displaced individuals’ integration and employment efforts has not been realized until now.
The Immigration Policy Lab at Stanford University in California and Switzerland’s ETH Zurich, in conjunction with Dartmouth College, has developed a data-driven algorithm to improve the process of placing refugees within a host nation.
When considering where to assign a new arrival, governments and resettlement agencies can use the algorithm to evaluate the outlooks of similarly situated refugees in terms of age, ethnic background, nation of origin and skill levels.
While the tool cannot determine nuanced reasons why a refugee struggles in a given location, it can detect systematic patterns that can be used to send refugees with similar characteristics and backgrounds to places where earlier refugees prospered.
“Algorithmic assignment holds the potential to simultaneously improve outcomes for refugees and the communities in which they are resettled,” said IPL affiliate Jeremy Ferwerda, an assistant professor of government at Dartmouth.
To train the tool for use in the United States, the team used data on more than 30,000 refugees aged 18 to 64 who were placed by a major resettlement agency between 2011 and 2016. The algorithm was then used to place refugees arriving at the end of 2016.
Compared to historical outcomes, the average refugee was more than twice as likely to find work if placed by the tool – an increase in employment likelihood from roughly 25 to 50 percent. Refugees’ expected employment levels spiked across the board, including for those who were considered most and least likely to find jobs.
The researchers’ trials also found that, had the algorithm been used, the average employment rate for displaced individuals throughout the U.S. would have been 41 percent higher.
When the tests were repeated for Switzerland, the improvements were even more stark.
Using data from the Swiss migration secretary, the team examined refugees who received subsidiary protection and were placed across 26 regions between 1999 and 2013.
After training the tool – which constantly mines data on refugee outcomes – on the earlier data, the team tested it on refugees who arrived in 2013. While their employment rate was 15 percent, it would have been 26 percent had they been placed in the algorithm-derived ideal location – a 73 percent increase.
“Our goal was to develop a tool that not only worked well but was also practical from a real-world implementation standpoint,” said IPL data scientist Kirk Bansak.
“By improving an existing process using existing data, our algorithm avoids the financial and administrative hurdles that can often impede other policy innovations.”