When asked by the Chronicle of Higher Education why the chain of for-profit schools he oversees had such a high three-year cohort default rate, Arthur Benjamin, the chief executive of ATI Career Training Center, cited the fact that his institutions did not provide loan counseling after two years. What’s funny about that explanation is that by attending a school that participates in the Federal Family Education Loan (FFEL) Program, ATI’s students should have been getting that counseling and assistance anyway from a set of entities known as guaranty agencies.

Besides the funding source, one of the biggest differences between the bank-based FFEL program and the U.S. Department of Education’s Direct Lending is that students with loans in the former are given default aversion assistance from guaranty agencies—public or not-for-profits that receive over $177 million each year in special subsidies for this purpose. By contrast, these supports are not provided in the Direct Loan Program.

An analysis of the three-year cohort default rates by loan program, however, shows that the default rates for students at schools in the FFEL Program are nearly universally higher than those for the Direct Loan Program. This echoes findings issued by the Department a few months ago, which showed that Direct Loan institutions had lower default rates than those in FFEL. That analysis showed that the two-year default rate for direct lending was 4.8 percent, compared with a mark of 7.2 percent for FFEL. This 2.6 percentage point gap not only continues under the three-year rate, but it actually increases to 4.7 percentage points, a result of Direct Lending’s default rate rising to 8.0 percent while FFEL’s goes up to 12.7 percent. It is also worth noting that this change represents a growth of 3.4 percentage points in Direct Lending’s rate, while FFEL’s grew 5.5 percentage points.

These findings are replicated whether the data is broken down by sector, tax status, or program length. In all but one instance, the default rate among institutions in the FFEL Program is not only higher than those in Direct Lending, but the increases in default rates that result from moving to the three-year measure is also larger.

The chart below shows the two- and three-year cohort default rates by program for different sectors and tax statuses.

As the data show, the only instance in which Direct Lending institutions performed worse than those in FFEL occurred at private, not-for-profit institutions of less-than two-years—the smallest sector included. In that situation, Direct Lending started with a default rate that was 1.2 percentage points above FFEL on the two-year mark, a difference that increased to 4.3 percentage points with the addition of the third year of measurement.

Defenders of FFEL like to point to for-profit institutions as the reason for the discrepancy in default rates. It is certainly true that the loan programs have pretty different default rates in that sector. Overall, Direct Lending’s default rate for the sector increased by 9.7 percentage points versus a 10.2 percentage point gain for FFEL. This means that the overall gap between the two is actually pretty small at 1.6 percentage points—less than the difference between the two programs at public institutions. That said, the largest Direct Lending-FFEL gap is the 6.3 percentage point difference between the three-year default rates at for-profit institutions of less-than two-years. This is an increase of 2.4 percentage points in the difference witnessed under the two-year calculation.

The most interesting story here, though, is what happens at public institutions. In that sector overall, there is a gap of 4.0 percentage points between Direct Lending and FFEL on the three-year measure, up from 2.5 percentage points with the two-year calculation. What makes this surprising is the difference at four-year publics. After similar not-for-profits, public schools of four-years or longer often have the lowest default rates among all institutions. As a result, one would expect their results to vary little by program. But in fact they do. The Direct Lending three-year default rate at these types of schools is 5.5 percent; for FFEL it is 8.0 percent. That’s a sizeable difference when discussing students that should be among the easiest to oversee and help keep in repayment.

On the other hand, the results at four-year not-for-profit institutions are closer to what one would expect. There, the difference is only about 1.0 percentage point between the two loan programs.  That is an increase of 0.5 percentage points from what it was under the two-year rate, but that still places it among the smallest differentials.

A Worthwhile Investment?

Defaulting on a student loan is a terrible outcome for borrowers. It wrecks credit, which can make it hard to get a job, buy a house, or do a number of other things. It also dramatically inflates the amount owed by as much as 25 percent thanks to collection fees. And this is to say nothing of the loss suffered by taxpayers who subsidize some portion of that debt. Therefore, it is clearly in everyone’s best interest to assist student borrowers in staying in repayment. We already try that in one program by providing tens of millions of dollars to guaranty agencies. But as the results show, the outcomes there are actually worse than the program that lacks this support, even among the students least likely to default. Maybe it’s time to try something else.

A note on methodology. The Department data does indicate what program schools are in, but large numbers of them are listed as “dual.” Instead, I labeled programs as being in FFEL, DL, or Both by drawing on the Department’s data center. I summed schools’ Stafford (subsidized and unsubsidized) borrowing from the 2006-07 academic year and labeled schools as FFEL if half their dollars were disbursed in that program. Schools with less than 10 percent of dollars in FFEL were labeled as DL. All others in between were marked as both. For schools that did not have a program listed here, I used the one provided by the three-year cohort default rate data. For those that did not have a program from that data source, I used a later year of loan volume data from the data center.  Likewise, sector classification was derived from the Integrated Postsecondary Education Data System.