Can Fintech Fix Student Lending?
Alt Underwriting, ISAs & Bootcamps
Hey all, Jason here.
With the coronavirus vaccine finally approved, there’s light at the end of the tunnel — things may ‘get back to normal,' but, realistically, not for another 6-12 months. In the meantime, the pandemic continues to rage — and, with it, varying degrees of quarantine restrictions and the accompanying economic pain.
Recession and heightened unemployment often lead to an uptick in university enrollment, as students seek to increase their skills or re-train for a new career. Which got me thinking about a $1.6 trillion problem: student loans.
Can Fintech Fix Student Lending?
Access to education is a prerequisite for social and economic mobility. But as higher education in the US has transitioned from a public to a private good, the cost has increasingly fallen on individuals. And as the cost has increased, so has the complexity of financing it.
While the COVID-era has seen a slight deleveraging in overall consumer credit, there’s one category of consumer debt that has continued to increase: student loans (which makes sense, given the accommodations granted by the CARES Act). Student loan debt has proven to be remarkably uncorrelated with other categories of consumer debt or the overall economy, steadily increasing in good times or bad:
The negative impacts of this ballooning debt -- on individuals and society at large -- are well documented:
With the average borrower making payments of $393/month, student debt crowds out other spending and is a drag on overall economic growth
Student loan obligations make it difficult for millennials and Gen Z to save, driving delayed home ownership, lower levels of wealth accumulation (vs. prior generations at same age), and greater financial fragility - disproportionately true for lower income, Black, and Hispanic populations
Damage to credit history from late payments, collections, and defaults, making access to credit more difficult and more expensive, resulting in higher financing costs, from credit cards to car loans and mortgages
Greater risk aversion, causing lower rates of entrepreneurship and small business formation
The fundamental factor driving student loan debt is one of price: it’s no secret that the price of college has increased dramatically (and much faster than underlying inflation).
While a bit outside the scope of this analysis, there are multiple trends driving the increase in the cost of education:
by far the most impactful is the dramatic decrease in state budget support for higher education (forcing universities to depend more on tuition revenue)
increased demand for higher education (and “credential inflation”)
increasing faculty numbers and cost
ancillary student services like luxury dorms, sports and recreation facilities, etc.
and a perception that price equates to quality
As costs have increased, higher education has transitioned from being a public good (paid for largely with tax expenditures) to a private one (paid for out of family income, savings -- or financed by borrowing).
The Government Approach: Finance It
As the cost of tuition at state- and private-run universities has ballooned, so has the complexity of federal aid to defray and finance the cost.
Depending on a student’s family’s situation, they may qualify for a mix of “need-based” and “non-need-based” aid, which may include a mix of loans, grants, work-study, and other programs. By far the largest portion of cost is covered via various loans.
The Federal Government as a Unique Kind of Lender
As a lender, the federal government is unique in a number of regards: its mandate isn’t generating a profit, and it can tolerate virtually unlimited losses.
The second part is what enables the first - the mandate is about “helping make college education possible”, regardless of credit risk. This has some good, bad, and just plain strange outcomes.
The combination of need- and non-need-based aid, including loans (regardless of borrowers’ ability to repay), has supported increased access to higher education.
If it’s not obvious at this point, the “bad” is the increasing mountain of student debt - much of which may never be repaid.
The two core principles of underwriting are assessing a borrower’s “ability to repay” (will they have enough free cash flow to make debt payments?) and “willingness to repay” (do they have a history of paying debts on time?).
With 18-year-old student borrowers, neither of these are possible.
Further, the mandate of the Department of Education is to expand access, not to pick “winners” and “losers” by looking at employment and financial outcomes for specific universities and majors.
The result isn’t pretty:
The burden is especially high for students who don’t complete their degree program but still must repay the loans they’ve taken — without the benefit of increased employment opportunities and income.
The ‘bad’ gets ‘worse’ in that, in almost all cases, student loans cannot be discharged in bankruptcy.
In a private market credit product, the interest rate a borrower is charged is linked to credit, inflation, liquidity, and interest rate risks.
For government student loans, the interest rates are set by Congress. For the most common “Stafford” loans, the interest rate currently is 2.75% for undergraduates and 4.30% for graduate students (excluding a 1.057% origination fee).
Because there is no up-front assessment of borrowers’ ability to pay, it’s not surprising that borrowers end up in unaffordable loans. Income-driven repayment typically caps a borrower’s payments at 10% of their disposable income (net income less some allowance for cost of living).
This sounds like a good enough solution on its face, but can result in situations where the monthly payment is less than the interest due, resulting in a loan with an increasing balance. While the balance will eventually be forgiven if on-time payments are made for 20-25 years, the result is another unanticipated consequence: a potentially hefty tax bill (as forgiven debt is treated as income).
What About Private Lenders?
While federal (public) loans account for ~92% of student debt, the ~8% of private loans still amounts to a mammoth $130 billion in outstanding balances.
Why is there a market for private student loans? While the aggregate (across years) loan limit for undergraduates varies, in many cases it is not enough to meet the rising cost of tuition and living expenses, forcing many families to resort to private borrowing. [Edit: there are a number of additional drivers of private student lending, including high-cost graduate school programs and for-profit undergraduate programs.]
Private student loans are made primarily by for-profit businesses like Sallie Mae, Discover, and Citizens Bank.
As for-profit institutions, they have a very different mandate (profit) and thus a different approach to underwriting (assessing ability to pay + willingness to pay).
Typical students seeking a private student loan will quickly experience this difference in underwriting. Because most student-aged applicants lack a credit history (and meaningful income), they are extremely unlikely to qualify on their own.
That’s why over 90% of private student loans are co-signed -- typically by a parent or family member. But, not all parents have the credit history or income to qualify as a co-signer. Perversely, this is disproportionately true for low income families, where the need for additional financing is often higher.
Fintech to the Resuce?
Around the same time the first crop of lending fintechs were taking off, a number of startups, most notably SoFI, were focusing on one aspect of student lending: refinancing. There can be real benefits for borrowers from refinancing existing student loans (public and private):
lower interest rate, especially vs. government loans, many of which were as high as 6.8% at the time
lower monthly payment (especially if borrowers refinance into a longer term)
consolidate multiple loan payments
Part of the premise is that these lenders can better underwrite by incorporating education data (university, degree, major) into their model.
The resulting credit and customer acquisition strategy at SoFI, for instance, was to target the most well-qualified borrowers (eg, Stanford computer science degree) and offer attractive refinance rates, making the determination that these applicants are lower risk.
Notably, refinance does nothing to expand access to education; it can reduce interest expense on loans after the fact, but this benefit tends to accrue to borrowers who already are higher income / higher credit score and thus deemed less risky.
A Tougher Challenge: In-Study Loans
A number of fintechs are trying to tackle the challenge of originating private student loans for students studying at typical 2- and 4-year universities. While most all make use of education data to some extent, the products/approaches broadly fall into two categories:
-credit underwritten loans, which may incorporate education data as a variable, but are assessing ability- and willingness-to-pay, and likely still require a co-signer. These look and operate much the same as private student lenders; the incorporation of education data may benefit students at selective universities or in certain majors (and may disadvantage students in certain majors, less selective universities, and community colleges.)
While the CFPB issued a “No Action Letter” to Upstart, conceptually providing safe harbor for use of these data points, Upstart and Wells Fargo have come under fire from the Student Borrower Protection Center, which argues the approach is tantamount to “educational redlining,” with students at Hispanic-Serving Institutions (HSIs), Historically Black Colleges and Universities (HBCUs), and community college students systematically charged higher rates. [edit: Upstart offers refinance loans, not in-study. Other fintechs, like SoFI and Ascent, do offer in-study private student loans.]
Earlier this month, Upstart announced an agreement with the NAACP Legal Defense and Education Fund and the Student Borrower Protection Center to collaborate on reviewing fair lending outcomes of Upstart’s approach.
-"merit” or “future income” based loans, which rather than looking at an applicants’ past to determine approval or requiring a co-signer, use variables like school, program, graduation date, major, and GPA to predict an applicant’s ability to pay.
This approach is modeling the likelihood an applicant finishes their program; the likelihood they find employment post-graduation; and estimates their income, other debt obligations, and free cash flow once employed.
While this approach can expand access by avoiding the co-signer requirements, depending on how educational inputs are used, it may result in the kinds of biases the Student Borrower Protection Center highlighted with Upstart.
A Different Approach Altogether: Bootcamps & ISAs
Improving employment prospects through expanded access to education doesn’t necessarily mean a traditional 4-year degree program. After all, if there’s something tech is good at, it’s “disruptive innovation”.
Enter the bootcamps, Silicon Valley’s take on the humble trade school. Instead of teaching welding or HVAC, students learn UX design, web development, digital marketing, or data science. The promise is a new, high-paying career in as little as 12 weeks (General Assembly) or 9 months (Lambda School).
But the programs don’t come cheap; a typical General Assembly course is $12,000, while Lambda School has a sticker price (if tuition is paid upfront) of $30,000 or $15,000, depending on location. And, because they’re not accredited 2- or 4-year degree granting programs, students are not eligible for traditional public or private student loans.
The solution? A relatively unique financing mechanism that has come to be strongly associated with “coding bootcamps” and deeply intertwined in their business model: the Income Share Agreement (ISA).
The concept is simple: instead of paying tuition in advance, you commit to paying a percent of your income after you complete the bootcamp and once your income has reached a minimum threshold ($40,000 for GA and $50,000 for Lambda School), with a cap on the total amount repayable.
There are some arguments to be made for the ISA structure:
not credit underwritten, meaning many who wouldn’t qualify for a private loan or would require a co-signer can qualify
in theory, better aligned outcomes.
Unlike for-profit diploma mills, with the ISA structure, bootcamps don’t get paid if graduates aren’t able to land relevant jobs earning above the income threshold. Lambda, for instance, allows students to defer payments if they’re laid off or income falls below the $50,000 threshold; if deferred for 60 months, the students ISA obligation is terminated.
Though bootcamps can sell or take an advance against ISAs, potentially obscuring this alignment.
Bootcamps have leaned heavily on ISAs as part of their marketing, making aggressive marketing claims around ‘not paying any tuition’ until you’re hired.
But the bootcamp-and-ISA model isn’t without problems:
quality of instruction and employment outcomes don’t always live up to expectations
ISA structure can be unaffordable. For example, Lambda School charges 17% once you begin making $50,000 (gross).
The $708 monthly ISA payment would work out to ~22% of net pay on a $50,000 salary in California; for comparison, income-driven repayment on federal loans is typically capped at 10% of discretionary income.
ISA can be an expensive financing mechanism vs. a traditional loan, depending on your post-bootcamp income.
ISAs are not loans and do not have an interest rate, making it difficult for students to compare them with other payment or financing options.
Because ISAs are not loans, they lack some regulatory disclosures and protections (eg TILA) traditional loans offer.
Can’t We Just Cancel All the Debt…?
The tech industry’s entrees in the sector - focused on alternative underwriting/financing approaches and bootcamps - are no doubt well intentioned.
And while they may yield better outcomes at the margins, ultimately, the largest problem is the cost of traditional 2- and 4-year degree programs more so than the financing mechanisms.
Improving the cost side of the equation requires coordinated state- and federal-level public policy to rein in runaway costs at universities and better fund them, making them less reliant on tuition dollars.
With the economic fallout of the pandemic continuing to reverberate, accommodations for federal student loan borrowers have been extended (by one month, anyway).
President-elect Biden has voiced support for canceling up to $10,000 in student loan debt per borrower, and the progressive wing of the party is pushing Biden and/or Congress to cancel up to $50,000.
While such debt relief would afford those who currently owe some relief, it would do nothing to address the structural forces that led to the debt in the first place, nor would it increase equitable access to education and the opportunities that come from it for prospective students.
Great article. This is clearly a complex problem with no easy solution. In my opinion, some of this is tied to the maximum amount of federal loans a student is allowed to take each year. When the government increases that number, many schools increase tuition in lock-step (whether the tuition increase is justified by true cost increases or not). It seems like schools are getting free money, while the students are on hook for the rest of their lives.
Another elephant in the room is the performance of these loans. Many of the loans are in securitization pools that claim high performance. However, many of the underlying loans are not performing, but instead are automatically put into deferment programs so they don't get reported as delinquent. This practice of overvaluing these assets cannot continue indefinitely.
If the cycle of increasing both the loan amounts and the tuition year after year doesn't stop, college will be out of the question for many in the next generation.
Thank you Jason, your articles are always well written and informative.
A few thoughts:
1. First, Congress should finally pass the Retirement Parity for Student Loans and allow employers to match or make student loan payments on a pre-tax basis, just as they can with 401K contributions. (A) Generally, recent grads have negative net worth due to student loans (which is one of the reasons student loans are not forgivable in BK, with hardship). (B) The rate you pay on a liability is equal to earning a risk free after tax yield on an investment. Even with a rate of 2.75% on government loans, this is greater than the after tax risk free rate. (C) So when you are in a negative net worth and the cost of liabilities exceeds the after tax return on investment, you should pay down your liabilities first and build retirement funds (401K) second.
2. Instead of forgiving student debt of say $10K or $50K, the government should give the debtor the option of receiving this same amount as a cash payment and allow the debtor to decide if they will use the funds to pay down their student debt, or use it for more productive means. Even considering #1 above, it still may make sense to use the money for say a down payment on a home, stabilizing occupancy costs, and building net worth through home equity.
3. Going forward, the US could consider the UK's plan for financing public university, where the student repays 9% of their income above ~$25,000 USD. Payments are deducted by the employer out of your paystub, dramatically reducing servicing costs. Payments are made until the earlier of: your balance is paid off; you are permanently disabled; you reach the age of 65. And importantly each spouse or partner in a household are treated independently, so if one stays home to raise kids and the other goes off to work, the staying home spouse or partner is not penalized. Payments are made in arrears in case there is a job loss or a reduction in pay, and there is also the ability of the borrower to recapture these payments given certain circumstances.
4. Finally, let's consider revamping the accreditation system in the US. If WFH and remote learning taught us anything, it is that online learning (and working) works! Allow students to take some classes online, and make it easier to receive credit at whatever school you decide to attend. There is no reason you can't complete a degree in under four years.
Thank you again