A new GPS faculty member is leveraging her expertise in economics and AI to identify those most in need of financial help through phone call data
Phone calls connect people near and far around the world. Now, this same technology has the potential to help researchers, governments and nonprofits measure levels of poverty, and from there, identify which people and areas are most in need of aid.
That’s one facet of the research being conducted by Emily Aiken, who joined the UC San Diego School of Global Policy and Strategy (GPS) as an assistant professor in January.
Aiken’s research is interdisciplinary: She’s interested in development economics and in data science and machine-learning tools, the latter of which she uses to study poverty alleviation in developing countries.
“I was super happy to be offered this position at UC San Diego, which lets me practice both sides of my research identity,” she said.
In addition to her role at GPS, she has been appointed to UC San Diego’s Halıcıoğlu Data Science Institute, meaning that she will also have a hand in developing the machine learning methods that power her work in economics.
A new meaning for ‘cellphone data’
Her dual research focus can be seen in one of her most ambitious research projects, in which she worked with the government of the West African nation of Togo and the nonprofit organization GiveDirectly to optimize the way in which financial assistance was allocated to those in need during the COVID-19 pandemic.
In ideal conditions, governments can use census data or other demographic statistics to figure out which communities to prioritize when giving out aid. But in Togo, the last census was taken in 2011, long out of date. Now, developments in machine learning have made it possible to get an idea of people’s income level through other data points, ones that are commonplace and don’t require an army of door-to-door canvassers.
Specifically, Aiken used satellite imagery and phone data to actually infer household level poverty, which Togo’s government then used to target cash transfers to the poorest households.
Through that project, Aiken and her colleagues found that using this mobile phone data was more effective at identifying those in need than simply inferring people’s income based on general criteria, like their occupation.
By looking at anonymous call logs, the researchers could see how long people talked on the phone, what other mobile phone services (like texts or mobile internet) they used, and whether they placed their calls in multiple locations. Then, they fed this data into a custom artificial intelligence algorithm, which used the information as proxies for their levels of poverty.
Aiken’s work has highlighted how academia, policymakers and NGOs must work together to solve complex, deeply-rooted issues like poverty.
In follow-up surveys, she and the other authors found that the small cash transfers did in fact help those who received aid, signalling that the phone usage data was an effective way for the Togo government to assist the right people.
“We have shown that we can measure and or target poverty at one point in time, but now the really exciting question now is, can you target poverty in real time?” she said.
Doing so, she explained, would allow policymakers and researchers to see whether their allocation of funds was successful and identify households moving into poverty as a result of economic or environmental crises.
Given that her research was conducted in collaboration with Togo’s government, Aiken is confident that her work will have a tangible impact on policy in many other developing countries, and she has carried out similar studies in two additional countries, Bangladesh and Malawi. In addition, she said her work has highlighted how collaboration between academia, policymakers and NGOs must work together to solve complex, deeply-rooted issues like poverty.
‘Exactly the students I want to be teaching’
Aiken has long had an interest in using technology to create real-world impact, starting from her undergraduate days at Harvard.
“Even when I did computer science for my bachelor’s degree, I always wanted to do work that applied to social science and helped improve the world,” she said.
This took a more concrete form in her graduate studies at UC Berkeley’s School of Information, where she received her doctorate.
Aiken said that she’s most excited about teaching students at GPS precisely because they share those same goals of improving the world by finding evidence-backed solutions.
“I’m really excited to get to teach, especially at GPS,” she said. “These are kind of exactly the students I want to be teaching: students who are thinking about careers in government or civil society, where they’ll be making important policy decisions using data science and machine learning methods.”
Aiken said she’s happy that she’ll be doing this teaching in her home state of California. However, this will be her first time living in San Diego, and she’s excited to have easy access to some of the best surfing spots in the country, as well as to the Sierra Nevadas, where she plans to keep up her hobbies of back-country skiing and trail running.
“Doing all this stuff outside is going to be awesome,” she said.
