New York Fed found precisely that: https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr733.pdf.
National Bureau of Economic Research other papers as well (http://www.nber.org/papers/w17827.pdf is one that’s openly available and specific to for-profit institutions; there’s another more general one behind a paywall as well). I was going to read the one from the New York Fed but a quick look at it disabused me of that notion so I’ll quote a synopsis from an article about it instead:
“The report’s findings show that of all three programs the Federal Direct Subsidized Loans generated a 65 cent-on-the-dollar increase on college tuition, while Pell Grants generated a 50 cent-on-the-dollar increase on college tuition. The Federal Direct Unsubsidized Loans had little to no effect on the price.”
Looks like the “Bennett hypothesis” is largely correct.
One odd thing from the Consumer Reports article–the numbers in the article and the infographic are different. The infographic shows average student debt just north of $37k while numbers in the article ($1.3T and 42M; some other articles use 43M which further supports my confusion about the discrepancy) turn into ~$31k or almost 20% smaller than the eye-catching infographic. Finally, one thing I find annoying about these articles is their use of average numbers instead of the median. If you look at something like the following, https://trends.collegeboard.org/student-aid/figures-tables/distribution-outstanding-education-debt-average-balance-2014
you’ll see that the average is severely inflated by the top 14% of borrowers with the top 4% of borrowers really crushing it with, presumedly, med or law school loans. Macalester (endowment $762M) has a helpful distribution as well: http://www.macalester.edu/financialaid/loans/studentloansatmac/distributionofdebt.html. Looking at the Macalester graph, it’s clear the long tail that starts at $50k should’ve gone someplace they could afford. Should public policy change on their account?