Research

How America Built a $1.7 Trillion Student Loan Balance

The student loan balance is a structural story, not a story about reckless teenagers. Who owes what, why it compounds, and why small balances default most.

By the Euphoria team · 2026-07-21 · 7 min read

The average defaulter is thirty-nine years old

In the first quarter of 2026, 2.6 million federal student loan borrowers entered default, on top of roughly a million the quarter before. The New York Fed looked at who they were and found that their average age was 38.9 years, up from 36.4 before the pandemic, and that more than three quarters of them had been current on their loans or had no payment due at all back in 2019.

That is not a story about eighteen year olds signing things they did not read. It is a story about people in their late thirties who had been handling the debt, then stopped. Their credit scores fell an average of 91 points between the third quarter of 2024 and the fourth quarter of 2025, from 567 to 476.

• $1.72T the federal student loan portfolio, principal plus interest, in early 2026
• 42.6M people holding at least one federal student loan
• 10.6% share of student loan balances 90 or more days past due

The balance those borrowers owe into is one of the largest categories of household debt in the country, running neck and neck with auto loans at about 9 percent of the total and trailing only mortgages. It got that way through a sequence of structural decisions, most of which had very little to do with individual choices.

Two different totals, and both are right

Start with a measurement problem, because you will meet both versions of this figure and they are not the same.

The New York Fed's Household Debt and Credit Report put outstanding student loan debt at $1.65 trillion in the second quarter of 2026, a slight decline from the quarter before. That figure comes from credit reports, so it includes private student loans alongside federal ones, and it reflects what servicers have reported.

The Department of Education's own portfolio data put the federal portfolio alone at about $1.72 trillion held by 42.6 million recipients, and that number explicitly includes outstanding interest as well as principal. Two sources, two definitions, roughly $70 billion apart, and neither is wrong. If you ever need to reconcile a debt statistic with another one, the first question is always whether accrued interest is inside the number.

The year the government became the lender

For most of the program's history, federal student loans were made by banks and guaranteed by the government under the Federal Family Education Loan program. If a borrower defaulted, the taxpayer covered the lender. In 2010 Congress ended new FFEL lending and moved everything to direct lending, in which the Treasury is the lender.

Watch what that did to the shape of the portfolio. In fiscal 2007 the government directly held $106.8 billion and guaranteed $401.9 billion sitting on bank balance sheets. By early 2026 direct loans stood at $1,562.9 billion while the FFEL book had run down to $158.3 billion as those old loans were repaid or consolidated.

The total grew for two reasons that are worth separating. The number of borrowers rose from 28.3 million in 2007 to 42.6 million now, an increase of about half. The amount each one owes roughly doubled: divide the portfolio by the number of recipients and the average balance went from about $18,200 to about $40,500. More people borrowed, and they borrowed considerably more each.

Why the balance grows even when borrowing slows

Here is the part that confuses people who expect the total to fall once annual borrowing flattens out. A loan balance is not a bucket that only empties. It has an inflow no one has to authorize.

Interest accrues daily on the outstanding principal. If a borrower's monthly payment is larger than the interest that accrued that month, the difference reduces principal and the balance falls. If the payment is smaller, the balance rises even though the borrower paid in full and on time. Income-driven repayment plans set the payment from income rather than from what is owed, which means a borrower with a large balance and a modest salary can make every required payment for years and owe more at the end than at the beginning.

A balance can grow while every single borrower does exactly what the paperwork asks of them.

Aggregate that across 42 million accounts and you get a portfolio with real momentum. Add the loans in deferment, in forbearance, and in school, on which interest is also accruing, and the total can rise in a year when new lending falls. This is also why the recent policy design matters: the Repayment Assistance Plan subsidizes the interest a borrower's payment does not cover, which switches off that particular growth mechanism for loans enrolled in it.

The median borrower is not the average borrower

The scary number in most coverage is an average, and averages are the wrong tool for a distribution this skewed. The Department of Education publishes the portfolio broken out by how much each borrower owes, and the shape is striking.

About 52 percent of borrowers owe less than $20,000, and together they hold roughly 12 percent of the dollars. At the other end, the 8.5 percent of borrowers who owe more than $100,000 hold about 42 percent of the balance. The median borrower sits close to the $20,000 line while the average balance is over $40,000, which is what a long right tail does to an average.

That tail is mostly graduate and professional education. Undergraduate borrowing has annual and lifetime caps. Graduate and professional programs historically allowed borrowing up to the full cost of attendance, which is how a single borrower reaches $200,000 or more. Just over a million borrowers are in that top bracket, and they account for about a fifth of the entire federal portfolio.

Default runs the other way

Now the fact that reorganizes everything. If default tracked balance size, the graduate borrowers with six-figure debts would be the crisis. They are not. Default concentrates among small balances.

The Department of Education's three-year cohort default rate, which follows borrowers who entered repayment in a given year, ran at 9.7 percent overall for the fiscal 2017 cohort. It was 7.8 percent for borrowers who attended four-year institutions, and 16.7 percent for those who attended two-year institutions, where balances are far smaller. Rates at less-than-two-year institutions were similar to the two-year figure.

The mechanism is completion, not size. A borrower who finishes a credential gets both the debt and the earnings that tend to come with the credential. A borrower who leaves partway through gets the debt alone. A $6,000 loan against no degree is a harder obligation than a $90,000 loan against a professional license, because repayment capacity is about income, not about the size of the number. This also explains the age profile of recent defaults: these are not fresh graduates, they are people whose earnings never caught up to a debt they took on years ago.

What the data cannot tell you yet

Three things are genuinely unsettled, and anyone presenting this topic as solved is overreaching.

• The default series is broken for recent years. The three-year cohort rate for the fiscal 2019 group reads 2.3 percent, which is not an improvement in borrower behavior. It is the payment pause suppressing defaults inside the measurement window. The most recent clean comparison is the 2017 cohort, which is now old data.
• The rules are mid-rewrite. Several income-driven plans are being retired no later than July 1, 2028, and loans disbursed on or after July 1, 2026 have access to only one income-driven plan. Any projection of the balance rests on repayment rules that do not exist yet.
• Nobody knows the recovery path for 3.6 million recent defaulters. Default triggers collection, damages credit for years, and can reduce access to further aid. Whether these borrowers rehabilitate their loans or stay in default will move the delinquency statistics more than new lending does.

Why this matters for you

The useful takeaway is not a number, it is a habit. Whenever someone hands you an aggregate, ask what the distribution under it looks like, whether the average is being carried by a small tail, and whether the outcome being measured actually correlates with the thing everyone assumes it does.

That habit is exactly what Euphoria's data lessons train, using live figures and real distributions, so that the next headline built on an average has to get past you first.