The most controversial change from the Department’s proposed version of the gainful employment regulation in the spring and the one released at the end of October is no longer judging programs based upon the percentage of borrowers that defaulted on their loans within three years of entering repayment. Removing this program cohort default rate got a lot of attention because it was the only measure that looked at all borrowers, regardless of whether they dropped out. In its absence programs will only be judged on the performance of the students who finished.
In the lead up to the final regulation’s publication, the program cohort default had come under a great deal of fire from both the for-profit sector and community colleges. While for-profits had not necessarily challenged the measure itself, they had demanded that passage of a cohort default rate be sufficient to absolve poor performance on the debt-to-earnings rate (the other accountability measure and the one that remains) or vice versa. Community colleges, meanwhile, had concerns less about the measure itself than its failure to reflect low borrowing rates and the Department’s ability to properly administer challenges and appeals to the rate. Recent decisions to tweak a similar version of the default rate that applies to institutions also had not inspired confidence in measure’s validity as implemented.
Removing default rates in gainful employment does have serious implications for a number of programs. Based upon data released in March, there were 394 programs that failed the program cohort default rate and passed the debt-to-earnings rate. ((The actual number of programs might be a bit lower because we did not correct the data to reflect the higher assumed interest rate on debt payments for certificate and associate degree programs.)) Those programs will now just show up as passing. Another 205 were in the zone on debt-to-earnings and failed the program cohort default rate and will now get more time to improve than they otherwise would. Finally, another 213 programs had not debt-to-earnings data due to an insufficient number of graduates but failed the program cohort default rate. These programs will now have no result.
The change is also not insignificant from a student perspective—the 812 programs that are now no longer failing had 191,666 borrowers in them, about 19 percent of the total included in the data. The chart below shows this information in greater detail. (UPDATE: THIS PARAGRAPH AND THE ONE BEFORE IT WERE EDITED FOR CLARITY.)
The issue with losing the program cohort default rate is less about the measure itself or the number of borrowers or programs that would have gotten in trouble with it. Realistically, the cohort default rate is an imperfect measure that is probably too easy to game and fails to address the more important question of whether students are actually repaying their loans, not just avoiding the worst possible consequence.
The problem with losing the program cohort default rate is that it picked up a separate and apparently distinct set of policy issues in vocational programs: “lottery ticket” programs where graduates seem to look OK but significant numbers of students never finish. It’s missing a solution for that policy problem that's the issue, not the measure used.
Nothing illustrates this problem more than the online associate degrees issued by the University of Phoenix. They have seven associate degree programs that all pass the annual debt-to-earnings measure, with earnings that range from $27,000 to $46,000. All of them fail the default rate test, with half of them having more than 40 percent of their borrowers default.
All you need to understand how a program could have such seemingly inconsistent results is to look more closely at the number of students captured in each measure. Those seven programs represent 10,803 graduates. But they also have 127,574 borrowers and 20,356 defaulters. That’s 1,800 borrowers and 188 defaulters for every 100 graduates.
This pattern is not unusual among the 5,724 programs with at least 15 completers and default rate data. The chart below shows the median defaulters and borrowers per 100 completers for programs based upon both their debt-to-earnings and program cohort default rate results.
In general, programs passing the program cohort default rate had fewer defaulters per 100 completers than those failing the measure. But the difference is particularly stark for the 421 programs that passed debt-to-earnings, with those that failed the default rate test having over a defaulter-per-100-completers ratio nearly 2.5 times higher than those passing both measures.
Interestingly, the table suggests that there isn’t much of a difference in performance on programs that pass the debt-to-earnings test when looking at their borrowers per completer. But this is largely a function of the fact that 70 percent of the programs in question are certificates. This matters because certificate completers are the other major group of students who default on their federal loans, suggesting either these credentials are not as good at guaranteeing a student will end up not struggling with their debts or their short-term nature means their completion rates are so high that they are not as useful a predictor of non-default.
The table below addresses this issue by breaking out the results for passing debt-to-earnings programs by level.
Seeing the results by level underscores just how different the results are for non-certificate programs. An associate degree program that passes both measures has 172 borrowers for every 100 completers. One that only passes the debt-to-earnings measure has 603. The results are similar, though not quite as stark at the bachelor’s degree level, with programs passing both measures having 147 borrowers per 100 completers versus 190 for those that fail the default rate test. ((Removing the University of Phoenix brings the associate degree results closer to the bachelor’s degree figures.))