As I noted earlier, one result of the new trial three-year cohort default rates are that all sectors have higher average default rates than they did under the two-year measurement window. But these rates also represent a summation of all schools in a given sector, obscuring those who either had very small gains or huge increases.

Overall, the average institution saw its cohort default rate increase by 5.6 percentage points when moving from the two to three year measurement. Compared to the prior two-year default rate this represents a 94.5 percent increase on average.

But the institutional-level story is quite different. Some schools saw little to no change in their default rate. Over 475 institutions saw their cohort default rate increase by less than 1 percentage point (a few others had small decreases because of adjustments to the number of students in designated as being in repayment). On the other end, there were 687 schools that had an increase of more than 10 percentage points in their cohort default rate. And 27 of those had gains of 20 percentage points or more.

The school that gets the distinction of having the largest percentage point increase is the International School of Health and Beauty in Lauderdale Lake, Fla. The institution looks pretty good when viewed under the two-year measure, posting a default rate of just 5.4 percent. But add an additional measurement year and the default rate jumps to 42.9 percent—a gain of 37.5 percentage points. Other especially bad actors include the Los Angeles-based International Career Development Center (37.2 percentage point gain) and Washington, D.C.’s own Bennett Career Institute (32.1 percentage point).

The International Career Development Center also earns the dubious distinction of having the largest percentage increase in its default. Under the two-year formula, just nine of its 1,125 borrowers defaulted—a rate of 0.8 percent. Within a year, 419 more borrowers had defaulted, increasing its rate by an astonishing 4,651 percent.

The chart below show the 10 largest percentage point increases among for-profit institutions.

Here is the same chart for public institutions.

And finally, for private, not-for-profit schools.

There are two immediate takeaways from this information. On the one hand, there are a number of institutions that had bad default rates under the old data and that figure only gets worse when viewed through a longer measurement window. This should certainly raise questions about why students at those schools are failing to pay back their debt at such incredibly high rate.

But the more interesting questions arise around the second group of schools. These institutions, such as the International Career Development Center, look like they are doing an excellent job preventing defaults on the two-year measure, recording rates that are well below average. Once the third year is introduced, however, their defaults skyrocket. At first glance this would certainly seem to indicate that the institution is engaging in some sort of gaming process—such as placing students in forbearance or giving them a short-term job—that is allowing them to shift all their defaults into the future. These significant default rate fluctuations should hopefully be attracting the attention of the U.S. Department of Education.

Love them or hate them, the trial three-year cohort default rates are the beginning of a new measure that in a few years will start holding colleges accountable for having lots of their former students default on their loans. But as the Department works to implement this measure, it should keep in mind the significant increases demonstrated by a number of these schools and take steps to ensure that institutions do not just find loopholes in the system to hide defaults and instead think of meaningful ways to assist borrowers. If the new metric forces real change, such as making colleges do more to help students complete, lower costs and debt levels, or provide a higher quality education, then the cohort default rate will have served its purpose. But if it instead becomes just another federal metric to undermine (think 90/10), then we will have ended up with little more than more defaults that are just lurking just behind the edges of the measurement window.