There but for the grace of God go I:
Back in January, the Urban Institute—a widely respected Washington think tank—published a statistic about student debt that pretty soon caught fire among economics journalists. Based on an analysis of federal data, its researchers concluded that 49 percent of all outstanding education loans belonged to highest-earning quarter of American households.
In other words, student debt seemed to be heavily concentrated among the upper middle class. The number soon took off after David Leonhardt cited it in a New York Times column arguing that popular left-wing proposals to simply cancel all of those loans in a giant jubilee would be extremely regressive. I cited it too in some of my work on forgiveness. And it’s popped up frequently in this week’s conversation about Sen. Elizabeth Warren’s plan to wipe out much of America’s student loan burden, including in a Friday Washington Post column by Catherine Rampell that argued the proposal would be a giveaway to high earners.
It turns out that statistic was also wrong. After a bit of poking around this week, I learned that it was the result of a coding error by Urban’s researchers. The correct stat is that the top quarter of earners owe about one-third of all U.S. student loan debt—a significantly slimmer share.
Here’s the reason:
It turns out that using aweights with Stata's collapse command produces the wrong answer for sums (pweights are the correct option for a survey dataset like SCF).
— Matt Chingos (@chingos) April 26, 2019