United States and Switzerland, 2011-2016

Human-Centered Algorithms for Refugee Employment

Big data for refugee employment

ACTORS

Stanford University Immigration Research Lab, ETH Zurich, Dartmouth University, refugees, refugee resettlement agencies

Solution

Researchers at Stanford University's Immigration Research Lab and ETH Zurich, helped by Dartmouth University, used machine learning to explore this problem, observing refugees and recording their job placement within 90 days of arrival. Machine learning provides the opportunity to use quantitative data, such as age and whether or not a given refugee speaks English, as well as qualitative data, which includes factors such as nationality and level of education that are harder to measure with number crunching. Machine learning also uses the capabilities ofto use big data to identify trends that may otherwise go unnoticed - for example, certain regions of a host country may have higher success rates than others, or refugees from French-speaking African countries may find more jobs more quickly in Switzerland than other refugee groups because they speak French.These data enabled the researchers to optimize refugee placement, increasing hireability by 40% in the United States and by 70% in Switzerland.

Peace Engineering Takeaway

The integration of engineering tools such as machine learning and big data analytics into existing refugee integration programs can provide critical insights and a better understanding of what works - and doesn't work - in those programs. These insights can improve these programs in the short-run, and lead to better program design, and ideally better peacebuilding outcomes, in the long-run.
Last updated January 2018. Powered by Webflow.