More than a numbers game: auditing algorithms through an economics lens

The impact of algorithm use

The most important questions about how we use algorithms – in sectors from criminal justice and health care, to education and hiring – arise from the tensions between human and machine decision making.

“How do we determine if decision makers, whether human or algorithms, are making decisions that are good, fair, or that take society's preferences into account?” says Daniel Björkegren, assistant professor of economics, another organizer and presenter at the conference. “This could shape policies around affirmative action, digital addiction and issues with social media, and whether and in what cases we should allow computers to make important decisions on behalf of people.”

Some algorithms are based on human decision making, but humans don’t always make the fairest or most effective decisions. While using algorithms based on large amounts of data might lead to better and more accurate decisions than relying on humans, who have a variety of behavioral biases, algorithms can end up reflecting those biases and producing outcomes that are discriminatory along gender or racial lines. 

Examples include who gets access to bank loans or mortgages based on their credit scores, employment websites using algorithms to determine who is visible to recruiters searching for specific job criteria, and “risk scores” used by judges to determine whether someone is eligible for bail after being arraigned.

“There are a lot of big societal questions about algorithms,” says Roth. “It would be hard for any one field to solve them. That’s why this is an area where folks in different fields really want to work with each other.”

Making collaboration possible

The workshop was sponsored by the Orlando Bravo Center for Economic Research, which was established at Brown in 2019 with a gift from the family foundation of Orlando Bravo ’92, P’26. A well-known private equity investor, Bravo aimed for his gift to expand research innovation, collaboration, and training for the next generation of economics scholars at Brown.

In the third year of the center’s existence, more than 30 working papers have been released through the Bravo Working Paper Series, and 19 research projects and 13 graduate students have received financial support.

“The Bravo funding has definitely been a springboard to more external funding from private foundations and government sources,” says Brian Knight, professor of economics, vice chair of the economics department, and member of the Bravo Center Steering Committee. “We’ve been able to support projects in macroeconomics, econometrics, and development economics, which requires data collection and field work.”

The Bravo Family Foundation also established two endowed faculty positions, which have contributed to the growth in the department’s faculty ranks.

“As the faculty grows, we feel we have more capacity to advise graduate students and involve undergraduates,” says Knight.

Both Hull and Björkegren incorporate discussion of algorithms into their courses.

“We want to get the next generation of scholars thinking about these topics,” says Hull. “We give them some idea of the open questions in the field, with the hope that they can push things beyond us when it comes to research as well as policymaking.”

Planning for the future

In the long term, the center hopes to make this workshop a regular event, with additional participants from the policymaking community.

“Bravo Center funding enabled us to get a critical mass of the top people in this field into the room,” says Hull. “Folks were talking in front of crowds that they otherwise wouldn’t have and meeting people they otherwise might not have. The potential is enormous.”