CFPB’s Math on Student Loans Ain’t Pretty

Rohit Chopra, “A closer look at the trillion,” CFPB.

I’m not a fan of the three-year cohort default rate as the metric for the vitality of the student loan program. It’s about as unhelpful as the unemployment rate, which can hide people who left the labor force, are underemployed, left the jurisdiction, etc. Just as the definition of “unemployed” is slippery (like, any job search activity in the previous four weeks), so too is “default,” which is no payments whatsoever over the previous 270 days. The CFPB has released, to my recollection, the first ever breakdown of federal student loans by repayment status (billions of dollars):

In-school Grace Repayment Deferment Forbearance Default Other Total
Direct 133.8 (24%) 40.4 (7%) 237.4 (42%) 75.6 (13%) 48.3 (8%) 30.5 (5%) 3.2 (1%) 569.2 (100%)
FFEL 12.2 (3%) 6.6 (2%) 256.3 (60%) 46.5 (11%) 42.8 (10%) 58.8 (14%) 6.3 (1%) 429.5 (100%)

I get 30 percent of all federal student loan dollars in deferment, forbearance, or default. Only 49 percent of the total are in active repayment. I’d love to see a comparison to credit cards, but aside from class concerns, I think it’d tell us that the federal loan program has been a spectacular, embarrassing failure.

Here’s the same thing for borrowers (millions):

In-school Grace Repayment Deferment Forbearance Default Other Total
Direct 7.9 (28%) 1.9 (7%) 10.8 (39%) 3.2 (12%) 1.8 (6%) 2.1 (8%) 0.1 (0%) 27.8 (100%)
FFEL 0.9 (4%) 0.5 (2%) 12.9 (56%) 2.3 (10%) 1.6 (7%) 4.4 (19%) 0.3 (1%) 22.9 (100%)

I’m not going to do the same calculation because there’s certain to be some overlap between DLP and FFELP borrowers, but I direct your attention to the FFELP’s default rate: 19 percent. That’s a better long-term indicator of where the DLP is headed, and remember, once you’re in default, there is no IBR. People in those circumstances will have to negotiate with their lenders, but my guess is that debtors who are old enough can beg for mercy from bankruptcy judges (and federal court judges dealing with the inevitable appeals).

Another interesting factoid is that the average balance by repayment status table (omitted) shows that the average amount owed by people who are in default is less than $15,000 for both DLP loans and FFELP loans. I expected the figure to be much higher based on the belief that people would default when their loan balances are hopelessly high. It’s probably weighted-down by people who borrowed a small amount of money to go to a for-profit and then dropped out or couldn’t find a decent-paying job.

The CFPB also gives a table of repayment plan choices for direct loan borrowers (not FFELP):

Outstanding balance
Billions of dollars
Millions of recipients
Average balance
Thousands of dollars
Standard 10-year plan 139.9 9.84 14.2
Plans based on income 72.3 1.58 45.8
Income-contingent repayment 20.1 0.63 31.9
Income-based repayment 50.9 0.91 55.9
Pay as you earn 1.3 0.04 32.5
Plans not based on income 107.4 3.35 32.1
Extended repayment 62.1 1.63 38.1
Graduated repayment 27.8 1.27 21.9
Extended graduated repayment 17.5 0.45 38.9
Other alternative repayment plan 4.4 0.23 19.1
Total of loans in these plans 324 15 21.6

Note that the average balance for the 910,000 people on IBR is $55,900. This is certainly a bellwether for law school debt and other graduate or professional school programs that rely heavily on Grad PLUS loans. Going forward, student loan debtors, including undergraduates, will be on the better-publicized PAYE, which will reduce the average amount borrowed.



  1. In sum, pre-K through grade 12 is designed to produce compliant workers. Why else would we want kids to sit for hours straight, and reading from safe, historically white-washed textbooks?

    “Higher education” is meant to make workers go into serious debt, so that they think twice about whistle blowing or causing trouble for their bosses or industry. When you owe a ton in student loans, you cannot rock the boat.

    1. Read the link Mossey:

      To learn more, we decided to analyze data released from the Department of Education, pulled from the National Student Loan Data System. We took a close look at the Federal Direct Loan and the Federal Family Educational Loan (FFEL) Programs. Federal loans also include the campus-based Perkins Loan program, which represent less than 1% of the total outstanding.

  2. 19% default rate on FFEL loans and yet the SLABS market is recovering fairly rapidly (85% of all FFEL loans were bundled into SLABS). It’s just insanity.

    1. Unemployed Northeastern, I’ve seen your comments elsewhere, and they’re excellent. You’re one of the few people who knows how the federal loan program works. I was wondering: When are you going to start your own blog?

      1. Thanks for the compliments, Matt. Whatever understanding I have of the student lending system is the result of years of reading the Chronicle of Higher Education, Inside Higher Ed, and other sites, and not being afraid to ask questions of some of the financial aid pros who haunt the comments sections of those sites. And some things are simple deductive reasoning: if a paper looks to good to be true (Simkovic-esque papers for the BA/BS *premium* have been pushed out fast and furious since 2008), look to see who funds it. Then look to see who funds them. Chances are, you will hit a for-profit student lender or college by then. Simkovic’s biases and obstreperous demeanor are well known to us by now, of course.

        Have you seen this study out of Demos last week? Basically, for a two-person household, each with college degrees and the average student loan balance ($53k), they will ultimately lose $208k in household wealth over their lifetimes versus their non-debt holding peers, primarily because of lower home equity and retirement savings (since that money is instead going to student loan interest & principal). And this is under a sunny-side-of-the-street scenario where each person works in a college-level job, never experiences unemployment, and has annual raises for their entire working lives. Imagine scaling those numbers up to two-lawyer households…

        I must say, you’re the third person this week to ask me if I am going to start a blog. Sadly, no. I focus most of my time trying to find one of those high wage premium jobs that I must inevitably get because I have a JD and a kid who washed out of both McKinsey and Davis Polk says that means I’m magic. There was a nice attorney listing north of Boston Monday for $10/hour, but it’s already been taken down. But I’m always looking…

      2. I hadn’t seen that Demos report. I’ll have to read through it. $208k? I think the median American household in either the 55-64 bracket or the 65 and up bracket have only $170k in net assets. That’s really bad.

      3. Just another reason why the exclusion of student loan debt from Simkovic’s study (and massive undercounting of law school debt when it is considered) thoroughly damns his conclusions. Well, that and all the other reasons (exclusion of the Classes of 08-13, blithe assumption that the profession will remain stable for 40 years as it falls at our feet, etc).

      4. And of course causation. Seeing a Harvard-trained law professor say with a straight face that signalling doesn’t matter in higher education outcomes was mind-blowing.

      5. Well, if you had gone to a better law school than Marquette, you would understand that alma mater signalling is statistically irrelevant in job outcomes!!!!

        End sarcasm

        P.S. You ever read Northwestern professor Lauren Rivera’s “Ivies, Extracurriculars, and Exclusion: How Elite Employers View Higher Education Credentials” in the journal Research in Social Stratification and Mobility? It’s a very candid, and wild, glimpse into the hiring proclivities of white-shoe law firms, investment banks, and management consultancies (McKinsey, Bain, BCG). The Chronicle of Higher Education summarized it thusly in their coverage: “Brown and Cornell Are Second-Tier.”

      6. The funny thing is that while I was reading Simkovic’s paper I imagined that if I walked into one of my Poly Sci grad courses with something like that my classmates and instructors would’ve given me no quarter whatsoever. It’s unbelievable that a bunch of “labor economists” would be routed by mere grad students—just like Reinhart and Rogoff.

        The irony is that Simkovic is now on the law prof market because he’s at Seton Hall without tenure. Talk about false consciousness, but I’m sure he’ll land on his feet and not have to put a picture on his LinkedIn profile or whatever.

        I’ll have to add Rivera’s paper to the reading list as well.

  3. Not only that – Simkovic graduated from HLS in… 2007. You know what that tells me? He has a gargantuan pile of private law school loans.

  4. While we’re piling on S/M, has anybody received a detailed SIPP breakdown of the careers that the 40% of JDs not ending up as lawyers (S/M footnote 7 or 8, I forget which) actually end up in – which (amazingly) still yield huge earning premia (despite “lawyers” already being something like the 5th or 6th highest paid BLS profession…).

    Knowing *that* would really help 500,000 or so JDs I can think of.

    Actually, I’m only half kidding – have S/M put up a SIPP data set (or is it online already?) that would let us figure out exactly where nearly half the damn “profession” actually goes?

    I’m assuming the SIPP data (at best) will provide some uselessly aggregated profession like “CEO” (“cure to poverty is wealth,” gee, thanks for the info…) but it would still be nice to extract some more granularity out of S/M.

    As it stands now, it looks like S/M are advocating the “underpants gnome”/black box/”the JD is *magic!* theory of career planning.

    Don’t just tell there is a premia, tell us where to find it!

    1. The SIPP data are free from the Census Bureau, and I’ve tried to look into them, but they need some kind of expensive extraction program to be intelligible. (Do you know how to access .dat files?) The best I got was a spreadsheet from the Census Bureau’s DataFerret that contained four thousand cells containing “46”.

      A footnote in Simkovic’s and McIntyre’s paper says that a preliminary analysis of those 40 percent is promising in terms of “wage premia,” but it’s certain to be 55-year-old CEOs averaged with 28-year-old retail clerks.

      You’re right about the authors not sharing the data. One of my other problems with their paper is that they didn’t tell us anything about the 1,300 J.D.s from the SIPP. How many are female? What are their ages? etc. They just found the data and then waterboarded it with natural logs of mean hourly wages.

      In the end, though, it doesn’t matter. The authors, like many “labor economists” apparently, assume that credentials cause higher wages, when in fact wages are determined by more mundane things like demand for the goods and services labor produces.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s