Department of Education (ED) officials announced Wednesday — the second day of the 2019 annual Federal Student Aid (FSA) Training Conference — that they reduced the number of FAFSA applications selected for verification by about 26% using new technology.
During the second general session of the day, ED’s Ed Pacchetti explained to thousands of financial aid professionals that FSA had begun using a machine learning model in October 2018 to select FAFSA filers for verification, which reduced the applications selected from 30% to 22% in 2019-20 — 900,000 fewer than in 2018-19.
Celebrating lower verification rate of 22% of filers while still controlling for improper payments. That's 900,000 fewer filers selected for verification for 19-20 compared to 18-19. Big thanks from @collegeaccess and our students. #fsatc2019— Kim Cook (@NCANCook) December 4, 2019
Craig Munier, the assistant director of policy liaison and implementation at FSA, said ED is constantly working to find a balance between reducing the verification burden on students and financial aid administrators while still ensuring that students are disbursed the correct amount of funds. Pacchetti explained that ED’s cost-benefit analysis showed verifying 22% of FAFSA applications resulted in the highest return in reducing improper payments, which ED has been under pressure to address.
“Only with the new technology of the machine learning model could FSA lower verification rates while controlling improper payments,” Pacchetti said. “... We hope that all of you have felt the reduction in burden this year.”
Pacchetti added that the new technology has consequently reduced the number of filers who never complete the FAFSA after being flagged for verification — known as “verification melt” — which is an issue NASFAA has been outspoken about. He said ED has allowed 100,000 more students to continue the process of accessing federal financial aid based on its estimates that verification melt affects 11% of selected filers, and added that using others’ estimates that verification melt affects closer to 25% of selected filers, that figure rises to 225,000 students.
Pacchetti said that moving forward, ED will continue to work toward reducing verification melt and improving its process for verification selection.
“The model that we are using is retrained every year, and it gets smarter every year,” he said, adding that ED is also rethinking its cost-benefit model for verification to include more costs.
Publication Date: 12/5/2019