In a post last week, I observed that the latest cohort default rate (CDR) data shows that students are failing to pay back their federal loans at the highest rate since 1998. When talking about the numbers a few weeks ago, Education Secretary Arne Duncan attributed the increase to continued economic problems, saying in a statement “This data confirms what we already know: that many students are struggling to pay back their student loans during very difficult economic times.”
But what if that’s not actually true? What if the rate increases are instead the result of a policy change in the mid-2000s that allowed lots of borrowers in the bank-based Federal Family Education Loan Program (FFELP) to distort the borrower cohorts? Here’s what one reader wrote to me about default rates over the weekend:
If you take out the impact of the temporary two year FFELP “in school” consolidation suddenly discovered by the last Administration (2005 and 2006), which jacked up denominators 30% artificially and thus artificially deflated the CDRs of 2004, 2005, and 2006 by moving borrowers “backwards” into the wrong denominators—preventing any chance of their default due to immediate long-term in-school deferment on the new consolidation loans—and leaving CDR 07, 08 and probably 09 and 10 cohorts filled with an over-representation of bad borrowers, the trend line would be a decrease through CDR 08 rather than an increase from CDR 06 to CDR 07 and from CDR 07 CDR 08.
The reader is referring to the “early repayment loophole,” which allowed borrowers in the FFEL Program to enter repayment while they were still in school. This provided a benefit to borrowers because the interest rate they paid on loans while still in school is lower than what they would have to pay after it entered repayment. At the same time, entering repayment early meant that borrowers would no longer receive a six-month grace period when they left school or graduated. (Because students can defer their loan payments while in school, borrowers who consolidated while still enrolled did not have to start making payments until they graduated or left.) Long story short, in exchange for giving up the six-month grace period, borrowers could consolidate their loans and lock in a lower repayment rate, and still avoid making payments while enrolled. (This dear colleague letter from 2005 explains the loophole in greater detail.)
But lenders benefited from this opportunity as well. Consolidation loans had a subsidy of 2.64 percent at the time, which was higher than what was given for loans in repayment (1.74 percent) or Stafford loans (2.34 percent). So getting a loan to consolidate early raised the subsidy by as much as 90 basis points. To put that in context, lenders acted as if they would go out of business thanks to a 55 basis point subsidy cut in 2007.
So what does that have to do with cohort default rates? Borrowers end up in a cohort based upon when their loans enter repayment. Typically, that was six-months after leaving school or graduating. But if a borrower entered repayment early, then they would enter repayment in an earlier cohort than they should have. Moreover, because their loans would receive an in-school deferment, they also couldn’t actually default on those loans until they left school, making it even harder for them to default within the two-year measurement window.
The reader explains the mechanics of this cohort shifting:
For example, a FFEL borrower whose Staffords have anticipated date of completion of 2007 (and anticipated repayment date of six months afterward, in 2008) performs an in-school (early conversion) FFEL consolidation in 2005.
The Stafford loan, or loans, is “put into repayment,” paid off, and given a repayment date equal to the paid-off date. The new consolidation loan is placed into an in-school deferment. Thus, the borrower has been moved “backwards” to the 2005 CDR denominator. While a default on the new consolidation loan would be linked back to the Stafford loan(s), this is impossible while the paid off loans are in an in-school deferment. While it is possible of course that some of the early conversion borrowers quickly dropped out, this would still give them quite an abbreviated timeframe to default and end up in the numerator, in comparison with the regular life of their loans.
You can see the effect of moving borrowers into earlier cohorts just by looking at the number of students tracked in each cohort over the past few years (see chart to the left). Unfortunately, I couldn’t find a good breakdown of borrowers by loan program for the earlier years, so the data here is for all borrowers. Had it been broken down by program, the effects in the FFEL Program would have been clearer.
As the bolded rows show, the 2004, 2005, and 2006 cohorts all saw much higher growth than all the other years, while the 2007 cohort actually shrunk a good bit. The department finally cut off the practice on July 1, 2006 so the effects are only visible in a few years.
The effects of the early consolidation loophole on the cohort matters even more because the risk of borrower default varies. Entering repayment early and locking in a better rate is a pretty sophisticated move, so it’s likely that the borrowers that took advantage of this policy were more financially savvy and would have defaulted at a lower rate. In effect then, this policy cherry-picked some of the best borrowers and moved them into an earlier cohort, as the reader noted “leaving CDR 07, 08 and probably 09 and 10 cohorts filled with an over-representation of bad borrowers.”
The chart to the right shows what the default rates might have looked like without the loophole in place. It’s hard to know exactly because we don’t have good statistics on the types of borrowers who took advantage of the policy. So the chart shows what the default rates would look like if the cohorts from 2004 through 2008 were arranged in increasing size. Again, without breaking the figures out by loan program, the effects are not quite as noticeable since the early consolidation loophole affected the cohorts of FFEL borrowers.
While this change would still show an increase in defaults over the 2005 figure, the story of regular increases doesn’t hold. In fact, the default rate for 2008 would actually have decreased a bit relative to the prior year.
Obviously this estimate can’t set the whole story straight. Most importantly, we don’t know how many borrowers in those 2004 through 2006 cohorts should have defaulted but didn’t thanks to the loophole. If those were properly accounted for then my guess is you would see higher rates for 2004 and 2005 that would make the overall trend look much flatter and with less of a clear pattern in one direction or the other.
I’ll close with the reader’s thoughts on this matter:
The “real” decline in CDRs illustrates a number of possibilities, which should then, in turn, be assessed in terms of analytical proof and likelihood:
(1) The American economy is much more resilient than the pundocracy has given it credit;
(2) Federal student loans are generally immune from the vagaries of the economy due to a wide variety of consumer protections unavailable on other credit products;
(3) The CDR is simply a poor measure not only of default rate itself but also of the impact of short-term and long-term economic trends upon default rates.
I suspect that the pro-consumer features available to FFEL and DL borrowers generally provide a significant delay of the impact (but not immunity from) from the economy upon student loan defaults, and that, if the economic problems continue much longer, then the impact would be shown on four-year or five-year CDRs, if they calculated them for recent and earlier cohorts for comparison. The federal student loan programs are insulated from economic forces but not immune from them.