Betsy DeVos wants defrauded borrowers to earn negative money to get full relief
Thousands of defrauded borrowers finally have a path to getting all their loans forgiven by Education Secretary Betsy DeVos. All they have to do is show a bunch of graduates from their program earned a negative income.
This Kafkaesque nightmare is the latest step in a long-running refusal by the Trump Administration to grant full relief to borrowers who attended the now defunct ITT Technical Institute or Corinthian Colleges and who were determined by the Obama Administration to have been harmed by lies and misstatements by those schools. Instead of granting total relief, the administration had pursued a system of partial forgiveness, the latest version of which they unveiled Tuesday night.
While the concept of partial relief is highly questionable, the formula used by the administration is a combination of bad statistics applied to federal data on earnings by program that probably aren’t even representative of the people seeking relief. The end result is an unnecessarily cruel system that will give borrowers just a fraction of necessary relief because they didn’t have negative earnings and deem individuals with minimum wage jobs that might not even be in the relevant field as “successes.”
Here’s how the formula works. The Department of Education only wants to grant full relief to what they deem an “outlier” program. They define that by looking at the median earnings of graduates from the program a borrower defense claimant attended and looking at how those earnings compare to the median earnings of all similar programs. In other words, if an accounting program’s graduates have earnings similar to the typical wages of other accounting programs then the borrowers must not have been harmed. In the Department’s logic, it’s as if you’re running a race and as long as you finish in the middle of the pack you’re good.
To find an outlier they use a concept from basic statistics when data follow a “normal distribution” (something that looks basically like a bell curve). For data like that, if you take the mean and subtract two standard deviations observations below that will represent the bottom 2.5 percent of the distribution. Here’s a good image of the idea.
The Department is claiming then that only programs in that supposed 2.5 percent — two standard deviations below the median — should get full relief. Anything below the median but above that lower bound gets 75, 50, or 25 percent relief depending on where they fall.
This whole approach would get an F in a high school statistics course (and unprintable reactions from a college professor). For starters, this outlier approach isn’t even the right way to determine if there is a statistically significant difference between the averages of two sets of observations. A common way to assess something like that would be a t-test.
Not only is the Department using the wrong mathematical approach, it’s applying the standard deviation concept incorrectly. The problem is that whole outlier concept is predicated on earnings looking something like a normal distribution. They almost certainly aren’t. It’s more likely that there are many programs around the mean and then some outlier programs with much higher earnings. Something that looks like this is going to produce a large standard deviation such that when you subtract it from the mean you’re going to get strange results.
The result is that when the Department subtracts two standard deviations from the median earnings figure, it’s creating absurd results, not identifying outliers. Take for example the diploma for business administration and management from Corinthian. The Department says the median earnings are $18,104, but two standard deviations is $20,654. That means the threshold for 100% relief is -$2,550. That’s right, under the Department’s formula, a misled borrower would only be eligible for full relief if they earned less than $0, an impossible number to achieve.
This math gets even crueler when you look at what students actually earned. The Department says the median earnings for that business administration program at Corinthian was $11,669. That’s roughly 30 hours a week for 50 weeks at the federal minimum wage. That suggests those earnings aren’t a function of someone necessarily working in business — they just have a job somewhere and it can only go so low thanks to the minimum wage. Yet because it’s not that far from the median, claimants from that program only get 25 percent of their earnings forgiven.
In effect, the Department of Education is punishing low-wage workers for working. Not only is the Department’s math bad, but focusing just on medians of programs is a poor idea in the first place when there is a range of earnings within these programs. Consider a program with 30 graduates, 16 of whom earn $50,000 and 14 of whom earn $5,000. The median is $50,000. So even though a good chunk of graduates have low earnings the program looks like its outcomes are satisfactory.
It gets worse. For one, these data involve comparing outcomes for graduates. But lots of students didn’t finish these programs. Some 2014 data from the Department suggests that ITT Technical Institute had roughly two times as many students enter repayment each year as it did graduates. These data don’t tell us anything about someone who didn’t finish. It’s also not clear if the earnings years tracked and the students involved even line up with the period the claimants attended.
The end result is a formula that layers bad statistics on top of worse methodology to take away thousands of dollars in help from misled borrowers simply because they had what is likely a minimum wage job. And it lays out results, like a requirement for negative earnings, that even the barest amount of due diligence would have identified.
Enough is enough. Betsy DeVos’s last attempt to write a partial relief formula got thrown out in court for violating the privacy act. Hopefully this one suffers the same fate for its mathematical illiteracy.
