I borrowed over $30,000 for college, and after many years of repayment, I am now officially (student loan) debt-free. By a bizarre twist of fate, much of my professional life has been devoted to studying financial aid programs like student loans. In this essay, I reflect back on how my student loan experience compares to that which my research indicates would constitute an ideal student loan system.
I begin by offering a brief refresher on the recent evolution of student loans in the United States. I then dive into advantages of income contingent repayment and explain why the slow drift in that direction has been a good thing. Next, I contend that student loans should have no loan guarantees for lenders, or interest rate subsidies or loan forgiveness for students, and that we’ve been moving in the wrong direction on these matters. I then make the case for annual and aggregate lending limits and discuss how this has been weakened for certain types of loans. Finally, I argue that both students and the country as a whole would benefit from continuous competition among private lenders.
The evolution of student loans: early 2000s through 2021
There have been two big trends in U.S. student lending over the past two decades: 1) the replacement of private lenders by the federal government, and 2) the gradual drift from traditional repayment to income-driven repayment.
I took out my first student loan in 1999 and my last student loan in 2007. During that time, there were two main student loan programs—the Federal Family Education Loan (FFEL) program and the Direct Loan (DL) program. The biggest difference between these programs was that in the FFEL program, private financial institutions were the lender, whereas in the DL program, the federal government itself was the lender. Students tended to prefer FFEL because many lenders would often offer a discount (e.g., 0.25% lower interest) if you signed up for automatic electronic payments. Thus, of the outstanding student loan debt in 2007, Americans borrowed about 80% through FFEL and the remaining 20% through DL.
During the recovery from the Great Recession, policymakers tried to find revenue sources to pay for the Affordable Care Act. They created one source by essentially replacing all FFEL lending with DL lending. We’ll get into some of the details of the FFEL program shortly, but for now, suffice it to say that a combination of unnecessary subsidies to FFEL lenders and assorted accounting gimmicks led Washington to believe that eliminating the program would save the government almost $90 billion over 10 years, money that could be used to help pay for Obamacare. The Obama administration therefore ended FFEL, and ever since 2010, all student loans have been made through DL, where the federal government is the lender.
The replacement of private lenders was quick, but the other big trend in student lending, the shift toward income driven repayment, has been more gradual. Income driven repayment (IDR) refers to the series of programs that offer alternative repayment plans in which the amount a student owes is determined by how much he currently earns, rather than how much he borrowed. Instead of making payments for a predetermined length of time like 10 years, everyone pays the same percentage of their income each month, which means high earners pay off their loans quickly, while low earners take longer. This type of lending was largely pioneered by Australia and New Zealand, and it has spread to many other countries including the UK and, increasingly, the U.S. In the U.S., the standard repayment plan would require students to make a fixed payment each month for 10 years. IDR plans instead set the payment amount based on the student’s discretionary income (discretionary income is typically defined as 150% of the poverty line).
One of the first American IDR programs was Income Contingent Repayment, introduced in 1994. It required students to pay 20% of discretionary income either until the loan was paid off or for 25 years, after which any remaining balance would be forgiven. The most recent IDR program is the Revised Pay as You Earn (REPAYE) program, introduced in 2015, under which students pay 10% of discretionary income either until the loan is repaid or for 20 years (25 if the student has graduate loans). The most generous IDR program is the Public Service Loan Forgiveness (PSLF) program, under which students get their loans forgiven after just 10 years of payments (see my recent piece on how the Biden Education Department just made PSLF even worse than it already was). The shift toward these alternative payment plans has been gradual but substantial. In 2013, 11% of borrowers were using an IDR plan, but by 2021, that number rose to 32%.
Have these two trends moved us toward a better student loan system? To answer this, I explore how they align with four key characteristics of an ideal student lending system.
Student loans should have income contingent repayment
The first key characteristic of an ideal student loan system is that it should have income contingent repayment.
Student loans help solve a problem—many young people would benefit from a college degree but do not have the income or wealth necessary to pay for school. Student loans provide these students with the funding they need, which then hopefully generates the higher income that allows them to repay their debt. But while the basic rationale of student loans is similar to that of other types of investment loans—borrow money now to make a profitable investment and then use some of the gains from the investment to repay the loan—there is a good case to be made that a traditional loan repayment schedule is not a good fit for student loans.
Traditional loans like mortgages or car loans set a predetermined dollar payment for a predetermined amount of time (e.g., a mortgage payment of $1,500 a month for 30 years). This is the default repayment plan for student loans as well, with fixed payments made for 10 years. However, there are good reasons to believe that this traditional loan structure doesn’t work well for student loans. New college graduates (or dropouts, who often have loans too) tend to be either just starting their careers or looking to switch careers. As such, they can have highly volatile earnings for several years, including spells of unemployment which make being able to pay fixed amounts on a set schedule challenging. With volatile earnings, students may default on loans due to short-term liquidity issues rather than long-term insolvency.
A better approach uses income contingent lending. An income contingent loan sets each payment as a percentage of income. High earners will repay their loans faster than low earners. Income contingent loans are far superior for student loans because they solve the problems of unaffordable payments and prevent default due to short-term income disturbances such as periods of unemployment. The various income driven repayment plans that are currently available are a type of income contingent lending.
When I was a student, most of my loans were through the Federal Family Education Loan (FFEL) program. For students struggling with repayment, there were income driven repayment programs available, but the default option was the traditional fixed-payment-for-10-years option. That is still the case today, but borrowers have many more choices of IDR programs, and those programs have much more attractive terms. Crucially, many more students are now utilizing these programs—almost a third of borrowers as of the second quarter of 2021.
Thus, we have made a lot of progress in the last several decades in moving from a model based on traditional loans to a model utilizing income contingent lending. This transition could be radically accelerated by setting one of the IDR programs as the default (ideally a newly designed one), allowing students to select the traditional payment schedule if they desire.
Student loans should have no loan guarantees for lenders or interest rate subsidies or loan forgiveness for students
The second key characteristic of an ideal student loan system is that there should be no loan guarantees, interest rate subsidies, or loan forgiveness.
One of capitalism’s crucial advantages is market pricing, which provides both information about profitable activities as well as an incentive to pursue them. But if those price signals are distorted, then decisions likewise will be distorted. Loan guarantees distort the prices offered by lenders, and interest rate subsidies and loan forgiveness distort the prices students are willing to pay.
These distorted prices can then lead to sub-optimal outcomes. For example, colleges that fail to educate their students can continue to enroll new cohorts of student-victims each year, with the students’ loans forgiven years later. And students can take out excessive debt that they are unlikely to be able to repay, counting on taxpayers to bail them out. For example, before being publicly shamed, the California Western School of Law advertised to students with a “STOP WASTING YOUR MONEY ON STUDENT LOAN PAYMENTS” campaign, saying their students could have more than $100,000 in loans forgiven.
In addition to distorting price signals, loan guarantees, interest rate subsidies, and loan forgiveness also impose high costs on taxpayers.
The first major cost for taxpayers were loan guarantees for loans like mine that were made through the FFEL program. Recall that FFEL relied on private lenders, but the federal government then offered a guarantee to the lender. Had I defaulted on my loans, the lender would have been paid 97% of my balance.
There is a legitimate argument that a loan guarantee encourages more lending on better terms for borrowers. But as mortgage lending leading up to the Great Recession and the history of FFEL make clear (FFEL lenders were sometimes caught bribing college officials to be placed on preferred lender lists), providing a guarantee to a lender all but assures that no due diligence will be performed by said lender. After all, they face little risk if they make a bad loan, and they will instead shovel as much money out the door as they can, leaving the taxpayer to pick up the tab when things fall apart.
Since the government is the lender now, there are no longer loan guarantees (for new loans), but if and when private lending returns, it is crucial that loan guarantees are not reintroduced.
Interest rate subsides are another burdensome cost to the government. Historically, Congress set interest rates, often below market rates, which can dramatically increase the cost of the loans to the government. Nicholas Barr calculated the cost of such a subsidy in England, which sets the interest rate equal to the inflation rate, and found that the “interest subsidy is expensive: for every £100 the government lends, between £30 and £35 is never repaid simply because of the interest subsidy.”
The interest rate on the last loan I just finished repaying was actually below the rate of inflation, around 1%, providing an even bigger taxpayer subsidy. While I appreciate the generous subsidy, offering highly subsidized interest rates to the well-off is not an appropriate use of taxpayer dollars. Fortunately, Congress has already provided the basic framework to eliminate interest rate subsidies. The Bipartisan Student Loan Certainty Act of 2013 tied student loan interest rates to the government’s cost of borrowing (the undergraduate interest rate is set at the 10-year Treasury Bill rate plus 2.05%.) As Susan Dynarski and Daniel Kreisman note, “Student loans are appropriate neither for raising revenue nor for subsidizing college,” so the value of 2.05% should be adjusted up or down to ensure that the government is neither making nor losing money on student loans.
Meanwhile, loan forgiveness is another costly and inappropriate feature of America’s student loan system. Income contingent payments already ensure that student loan payments are affordable, and they provide that assurance for the student’s entire lifetime. In other words, income contingent loans completely solve the problem of unaffordable student loan debt. They also by definition include de facto loan forgiveness for those with earnings too low to repay their debt over their lifetime. Given that payments will always be affordable and that there is already loan forgiveness built into the foundations of an income contingent loan, it is bizarre that virtually every income driven repayment plan in this country also includes a time-based loan forgiveness feature.
Unfortunately, we’ve been moving in the wrong direction regarding loan forgiveness. The first income contingent repayment plan offered forgiveness after 25 years of payments, whereas today, some borrowers can get their loans forgiven after as few as 10 years.
Since income contingent lending already has de facto loan forgiveness built in, we should eliminate any time-based forgiveness for these programs.
For my student loans, the government offered loan guarantees and extremely generous interest rate subsidies, but little forgiveness. Today, since there are no private lenders, there is no longer a loan guarantee, and interest rate subsidies have been largely curtailed. However, loan forgiveness has become much more common and is quickly becoming a windfall for some. Borrowers in the Public Service Loan Forgiveness program, for example, have an average of almost $83,000 in debt forgiven.
To improve student loans for future students, we should 1) ensure that when private lenders are involved, there is no loan guarantee, 2) offer no interest rate subsidies to students or lenders, and 3) offer no additional loan forgiveness (beyond the forgiveness already provided by income contingent lending).
Student loans should have annual and aggregate caps
A third key characteristic of an ideal student loan system is annual and aggregate caps on borrowing.
There is considerable (and growing) evidence that student loans lead to higher tuition as colleges raise prices to exploit students’ increased ability to pay. Known as the Bennett Hypothesis, this phenomenon is a behavioral response to a statutory relationship. The statutory relationship refers to the way federal financial aid eligibility is calculated under current law. If a college raises tuition by $1, the student becomes eligible for $1 more in aid. The behavioral response kicks in when colleges respond strategically to this statutory relationship, noting that they can increase prices without substantially reducing students’ ability to pay, since the students just get more aid.
Because the Bennett Hypothesis is a behavioral response to a statutory relationship, you can fight it at either the behavioral response level or the statutory relationship level. The best way to amend the statutory relationship would be to use the median cost of college when determining aid eligibility rather than letting each college largely determine aid eligibility itself. If aid eligibility is determined by the median cost of college, then when a college raises tuition, its students no longer automatically qualify for more aid, thus breaking the statutory relationship that drives the Bennett Hypothesis.
Another method of limiting the damage from the Bennett Hypothesis that is already in use for some programs is an annual and aggregate cap on aid. For example, dependent undergraduates can’t borrow more than $7,500 per year and $31,000 over their lifetime.
Unfortunately, some borrowers aren’t protected by such caps. For example, graduate students and parents can borrow up to the full cost of attendance, which each college sets as it pleases. Thus, if the college raises tuition, these students and parents often just borrow more.
Since I took out loans, the loan limits for undergraduates were raised, which likely led to higher tuition. In addition, in 2005, the federal government introduced cap-free Grad PLUS loans for graduate students. To avoid letting student loans simply fuel tuition increases, all loans going forward should be subject to annual and aggregate limits.
Student loans should exploit continuous competition among private lenders
A fourth key characteristic of an ideal student loan system is competition among private lenders.
The federal government in the only lender for student loans, and this is quite unusual. Some argue that this is necessary because students lack collateral. For a mortgage or a car loan, the underlying asset being financed can be used as collateral for the loan. If you stop making mortgage payments, your lender can repossess your house and recover most of its losses. But with traditional student loans, there is no collateral, as there is nothing for the lender to repossess. Without collateral, lenders would charge a very high interest rate. The argument is that government-as-lender solves this problem, largely by ignoring the lack of collateral and charging an interest rate similar to the rate for collateralized loans.
Income contingent lending can also solve this problem since it uses the student’s future income as collateral for the loan. Income contingent lending can also utilize private lenders, and there is much to be gained from unleashing private lending. A recent study of mine details many of the advantages of having a competitive market in private lending, but the biggest are a reduction in malinvestment and more informed decision making.
The current government-as-lender system results in too much malinvestment. In a forthcoming study, I estimate that more than 100,000 students with loans graduate from a program that fails a debt-to-earnings test each year, meaning that their students are highly unlikely to be able to repay their loans. The federal government has simply proven incapable of restricting aid for programs that consistently lead to bad outcomes for students, financing such malinvestment year after year. Private lenders would not provide financing for these types of programs for the simple reason that they would lose money.
Private lending would also result in more informed decision making. Right now, a stellar student attending a top school and majoring in a high-demand field gets the same loan terms as a slacker student attending a college in name only and majoring in an unmarketable field. With private lending, we would see differential pricing as opposed to the federal government’s uniform pricing. This would benefit students by sending them signals about which educational investments are high-risk. But it would also benefit society more broadly by encouraging more students to enter high-demand fields and discouraging students from entering low-demand fields.
The FFEL loans that I took out did have private lenders, but don’t mistake that for a competitive market in lending. The government determined which students were eligible for loans, the amount of each loan, set the interest rates on the loan, guaranteed the lender an annual profit (if the interest rate set by Congress was low, the lenders would get payments to compensate for the low rates), and guaranteed the loan in the case of default. In other words, FFEL lenders determined neither their customers nor the price charged, faced virtually no downside even if the loan defaulted, and were subsidized by the government to ensure they earned a profit on the loan. This is not market-based lending—it’s cronyism. The main competition among lenders in such a rigged system is who can pay the most bribes. It is also why I wrote way back in 2009 that “FFEL should be taken out back and killed with a shovel.”
A competitive lending market looks like the mortgage market, not FFEL. What protects borrowers in the mortgage market is the competition from other lenders who will offer better terms if a financial institution tries to take advantage of a customer. We would benefit from such competition for student loans, too.
Overall, student loans have evolved in good and bad directions since I was a student. On the bright side, we are moving toward an income contingent system, which is a dramatic improvement. Inappropriate interest rate subsidies are also less of a concern. At the same time, loan forgiveness has gotten more and more generous over time, and many progressives are pushing for total forgiveness. Loans for graduate students and parents are not capped, fueling tuition increases. With any luck, by the time my kids are taking out student loans, we will have designed a better system.