Science & Research

Research Spotlight: Björkegren uses cell phone data to uncover economic trends

By
Science & Research Editor
Tuesday, October 20, 2015

This profile is part of a series focused on Brown faculty members and students engaged in science and research, with the purpose of highlighting and making more accessible the work being pursued at all levels across disciplines.

For all researchers, including economists, a paper is only as good as the data on which it is based.

In the United States, researchers have access to plenty of economic data that is collected by the government and other organizations. But in many developing countries, economic data is scarce, especially for rural populations.

Assistant Professor of Economics Daniel Björkegren uses data from people’s cell phones to bridge this data gap and study developing economies. Currently, Björkegren is working on two projects that link cell phone data to economic trends.

The first project looks at the adoption of cell phones in Rwanda from 2005 to 2009. Expansion of cell phone usage is a classic example of the network effect — as a product gains more users, it becomes more valuable.

When cellular service is implemented in developing countries, service begins in wealthier urban areas and later expands to rural areas, he said.

Björkegren examined how to make cell phone coverage in Rwanda socially optimal, rather than just profitable.

One method to achieve this goal is to mandate that a certain percentage of the population be covered by the carrier. Another method is to require the carrier to use some funds to provide free handsets to populations in rural areas, thus  boosting the network effect. Both methods increase social welfare, Björkegren said.

Rwanda was a good candidate for the study because it has only one major cellular carrier and the carrier was willing to provide anonymous cellular metadata, he said. Cell phone metadata includes the type of phone, the length of the call, the time of the call, money left on prepaid phone cards and the rough location of the caller based on information from the cellular tower.

“What can we learn now that we have this information about people’s behavior?” Björkegren asked. “Cell phones have spread and become part of people’s lives.”

In the second project, Björkegren analyzes cell phone data to predict loan repayment rates in an undisclosed Caribbean nation. Cell phone records from the country are linked to loan records. The identity of the borrowers was kept secret.

Cell phone consumers who left a minimum balance on their prepaid phone cards were more likely to repay loans, as were those who travelled more widely across the country, Björkegren said.

A third factor, periodicity — the variation in time and occurance of cell phone usage — also correlated with reliability of loan customers, Björkegren said, adding that this finding was unexpected.

The research was completed in partnership with the Entrepreneurial Finance Lab, a credit scoring organization for those with little or no credit. Björkegren said cell phone data could potentially be used to determine lending decisions for individuals.

Björkegren said he hopes to determine through future work whether the repayment factors can be generalized to other regions and to use this data to influence decisions about loans.