The official student loan default rate is 0.0 percent. Nine million borrowers are in default

The United States government publishes an official student loan default rate. Calculated on 3 August 2025 and released the following month, it is 0.0 percent.

It is not a typo, it is not stale, and it is not disputed. It is the number that decides whether a college keeps access to federal student aid, and it has read 0.0 percent for three years running.

On the same day the Department of Education published that figure, its own portfolio file was counting several million people whose loans had gone past a year without payment. By 31 March 2026, the most recent date the Department has published, that count reached 9.0 million.

Four numbers are in circulation. Every one of them is government output. They do not agree because they were never measuring the same thing, and the reasons are written down in a regulation, a statute, and a footnote at the bottom of a spreadsheet.

What is the student loan default rate?

There are four published answers. The official cohort default rate is 0.0 percent. The Department's portfolio file shows 9.0 million recipients in default, about one in five federal borrowers. The New York Fed puts 10.6 percent of student loan balances at 90 or more days past due. The flow of balances into serious delinquency is 7.83 percent, and it is falling.

Two of these rise while a third falls and a fourth sits at zero. That is not a contradiction to be resolved by picking the honest one. It is four different questions with four different answers, and the useful work is saying which question each one answers.

This publication has run into the same shape before. The credit card delinquency rate is 13.12 percent and 2.92 percent at the same moment, because one measure retains charged-off balances and a supervisory rule empties the other at 180 days. The IRS audit rate is a partial count that rises about 45 percent after it is first published. Student loans are the most extreme case of the pattern yet found, because here the spread between two official readings is not a factor of four. It is a factor of several thousand.

Why is the official default rate 0.0 percent?

Because the cohort default rate measures a two-year window that fell almost entirely inside the payment pause and the on-ramp that followed it. For the fiscal 2022 cohort, 112 borrowers out of 3,529,089 defaulted inside that window. That is one in 31,510, and the rate rounds to zero at every published decimal place.

The measure is old and narrow by design. Under 34 CFR 668.202, the cohort for a fiscal year is the set of students who entered repayment in that year, and a borrower counts as in default only if the failure to pay persists for 360 days before the end of the second fiscal year following. The fiscal 2022 cohort therefore consists of borrowers who entered repayment between 1 October 2021 and 30 September 2022, with defaults counted through 30 September 2024.

Now put the calendar next to it. Federal payments were suspended from 13 March 2020. Borrowers re-entered repayment on 1 September 2023. A twelve-month on-ramp then ran to 30 September 2024, during which a missed payment did not carry its usual consequences and loans were moved into forbearance after three missed payments rather than proceeding toward default. The cohort window closed on the day the on-ramp ended.

So the fiscal 2022 rate is not a measurement that came out low. It is a measurement of a period in which the mechanism it counts had been switched off.

The full series shows how sharp the break is.

The counts behind the last four readings are 91,475 defaulted borrowers for fiscal 2019, then 196, then 158, then 112. A statistic that identified 91,475 people in 2019 identified 112 three years later.

Does the zero mean anything for the schools it governs?

No. The cohort rate exists to trigger a sanction, and at present it cannot. A school loses Direct Loan eligibility above 40 percent in one year, or Direct Loan and Pell eligibility after three consecutive years at 30 percent or more. In the published file of 4,635 eligible schools, not one reached either threshold.

Those thresholds sit in 34 CFR 668.206. To see how far the current readings sit from them, the school-level file the Department publishes alongside the national briefing was aggregated for this piece. Summing the fiscal 2022 columns across all 4,635 rows returns 112 defaulted borrowers, which matches the Department's published national numerator exactly. That match is the check that the aggregation is reading the file correctly, and it licenses the rest.

What the rest shows: 100 schools out of 4,635 recorded even one defaulted borrower. The other 4,535 recorded none. The single highest rate anywhere in the file is 20.0 percent, and it consists of one defaulted borrower in a cohort of five.

There is also a structural feature of the rate that survives the pause and will outlast it. Under 20 U.S.C. 1085(m)(2)(A), a student who borrowed at more than one school is attributed to each of them. The same default can be counted against three institutions. The rate was never a count of people and does not claim to be.

Is the government counting default at 270 days or 360?

Both, in different files. The regulation defines default on a Direct Loan as a failure to pay that "persists for 270 days". The Department's portfolio reporting counts a loan as defaulted at 361 days. The Department states the choice plainly in a note to its own workbook, and the gap is about half a million recipients.

The regulation is 34 CFR 685.102(b), and its wording is not ambiguous. The reporting convention is note 62 of the Department's delinquency workbook: "While technical default is 271 days delinquent, default is defined as 361 days delinquent for reporting purposes to ensure consistency with Federal Family Education Loans (FFEL) reporting." The status definitions tab of the companion portfolio file says the same thing from the other side, defining the default column as loans "more than 360 days delinquent".

That is a candid footnote and there is no concealment in it. The consequence is arithmetic. On 31 March 2026, the federally managed portfolio held 520,000 recipients in the 271 to 360 day band. Every one of them is in default under the regulation. None of them is in the number.

The band is also a queue, and watching it drain is the clearest available picture of what happened over the winter.

The pipeline emptied into the stock. In two quarters the default count rose 3.8 million, or 73 percent, and the balance rose $103.0bn, or 88 percent. The 181 to 270 day band held 870,000 recipients at the end of March, which is the next tranche.

Are 9.5 million people in default?

The count is of loans, not people, and the Department says so twice. The status definitions tab warns that "recipient counts are based at the loan level. As a result, recipients may be counted multiple times across varying loan statuses." A separate note defines a recipient as the student who benefits from the loan, which on a parent PLUS loan is the child rather than the parent who signed.

Both cautions are load-bearing. Add the seven status columns for the federally managed portfolio at 31 March 2026 and they total 45.5 million recipients, against 42.6 million unduplicated recipients in the Department's own summary file. The excess of 2.9 million is the double counting the note describes: a borrower with one loan in repayment and another in default appears in both places.

The parent PLUS point is stranger and less known. There were $117.9bn of parent PLUS loans outstanding at the end of March across 3.5 million recipients. Where such a loan is in default, the person who owes the money is the parent, and the person the file counts is the child, who may have no obligation on the debt at all. A count assembled on that definition is a defensible administrative statistic. It is not a headcount of Americans in default, and every sentence that presents it as one is adding a claim the file does not make.

What can be said with the file as it stands: 9.0 million recipients and $220.3bn of principal and interest were in the default column on 31 March 2026, an average of $24,478 apiece, which is 12.8 percent of the $1,723.9bn federal portfolio. Against 42.6 million unduplicated recipients that is 21.1 percent, which is where "one in five" comes from. Against the 17.2 million recipients in active repayment, it is 34.4 percent. Pick your denominator and the same file will support one in five or one in three.

One more caution about vintage. The Department's portfolio files currently stop at 31 March 2026. Any figure describing a later date is not from them.

Which loans can never appear in the official rate?

Every PLUS loan. The cohort is defined by statute as students entering repayment on Stafford, SLS, Direct Subsidized and Direct Unsubsidized loans, plus the portion of a consolidation loan that repaid those. Grad PLUS and parent PLUS are absent from the list. That is $248.2bn, or 14.4 percent of the federal portfolio, outside the measure entirely.

The exclusion sits in both layers of the law. Section 1085(m)(1)(A) names sections 1078, 1078-1 and 1078-8 and the consolidation portion at 1078-3, and does not name 1078-2, which is the PLUS program. The regulation at 668.202(b)(1) repeats the same list in plain terms.

An unquantified further share of the $529.7bn consolidation book is outside the rate too, because a consolidation loan enters the cohort only to the extent that it repaid an included loan. The Department does not publish the split.

There is a second gap on the FFEL side. Under 668.202(c)(1)(i), a FFEL borrower is not treated as in default for cohort purposes "unless a claim for insurance has been paid on the loan by a guaranty agency or by us". What is being measured there is the completion of a claims process, not the conduct of a borrower.

None of this is hidden and none of it is recent. It does mean that the one default statistic with legal force is blind to the fastest-growing and highest-balance segment of the portfolio, and blind by design.

Why is the New York Fed's number falling while the Department's is rising?

Because the two measure opposite things. The Department's 9.0 million is a stock, and a stock only grows until borrowers exit through rehabilitation, consolidation or repayment. The New York Fed's 7.83 percent is a flow, the balances newly transitioning into serious delinquency, and a flow falls as soon as the wave passes.

Both series are in the New York Fed's report for the second quarter of 2026, released this month. Its balance measure is rising: 10.6 percent of student loan balances were 90 or more days past due, against 10.3 percent in the first quarter and 9.4 percent a year earlier. Its flow measure peaked at 16.19 percent at the end of 2025, read 10.86 percent in the first quarter and 7.83 percent in the second, which is below the 12.88 percent recorded a year ago.

Two cautions travel with that 7.83 percent, and neither is usually attached to it. The chart note says the series is a four-quarter moving sum, so it is not the share of balances that went bad last quarter. And the report's own by-age chart puts the all-borrower line at 7.44 percent for the same quarter, on a different sample, which is a reminder that even inside one publication the number has a construction rather than a value.

The age breakdown is the part of that chart worth reading. At 31 March 2026 the transition rate rises monotonically with age: 5.32 percent for borrowers aged 18 to 29, 6.49 percent for 30 to 39, 8.05 percent for 40 to 49, and 9.86 percent for 50 and over. The population falling behind fastest is the one furthest from graduation.

The universes differ as well. The New York Fed measures $1.65tn of student debt from Equifax credit records, including private lenders, while the Department counts $1.72tn of federal loans in its own system. The credit bureau total for all student debt is smaller than the federal total alone. Two institutions, two universes, two numbers for the same borrowing.

What does default actually do?

It moves the debt into collection. The Secretary or a guaranty agency may garnish up to 15 percent of disposable pay without a court judgment, and the debt can be recovered from federal payments. The published exit is rehabilitation: nine payments in ten months, available once.

The garnishment ceiling is 20 U.S.C. 1095a(a)(1), which caps the deduction at 15 percent of disposable pay, permits more only with the borrower's written consent, and requires 30 days of written notice before proceedings begin. Rehabilitation is 34 CFR 685.211(f): nine voluntary payments, each within 20 days of its due date, across ten consecutive months, with the amount set from the borrower's total financial circumstances and floored at $5 until 1 July 2027 and $10 after that date.

That floor is the mechanism by which the stock can shrink without the underlying balances being paid, and it is also why a rising default count and a falling flow can coexist for several quarters.

Where this reading could be wrong

The weakest link is vintage. The Department's portfolio files were last refreshed for the quarter ending 31 March 2026, and the default count has been moving by more than a million recipients a quarter. The June quarter will change the level materially, and possibly the direction, since the 271 to 360 day queue feeding the stock has drained from 3.62 million to 0.52 million in six months. A stock that grew because a backlog cleared does not keep growing at that rate once the backlog is gone.

The second caution is about the cohort rate. Nothing here says the measure is defective. It says the measure answers a narrow question, about one year's entering class, on a 360-day test, over a two-year window, for a subset of loan types. The zero readings are a correct answer to that question during a period when repayment was suspended. The fiscal 2023 cohort, whose window closes on 30 September 2025 and whose rate should publish in the autumn of 2026, is the first that will contain a full stretch of ordinary repayment. That is the number to watch, and it is the one that will tell whether the pre-pandemic 10 percent range returns.

Third, the school-level aggregation in this piece covers fiscal 2022 only. The same file carries trailing columns for fiscal 2021 and 2020 whose totals do not reconcile with the Department's published briefings for those years, and rather than guess at the reason, this piece uses the briefing figures for every year except the one it could verify by matching the numerator exactly.

Frequently asked questions

What is the student loan default rate in 2026? It depends which measure is wanted. The official cohort default rate for the fiscal 2022 cohort is 0.0 percent. The Department of Education counted 9.0 million recipients in default on 31 March 2026, about one in five federal borrowers. The New York Fed put 10.6 percent of student loan balances at 90 or more days past due in the second quarter of 2026.

How can the official rate be zero when millions are in default? The cohort default rate counts only borrowers who entered repayment in one fiscal year and defaulted within roughly two years of doing so. The window for the most recent published cohort ran from October 2021 to September 2024, almost all of which fell inside the payment pause and the on-ramp that followed, when defaults on federally held loans were not occurring.

When is a federal student loan in default? At 270 days of non-payment under 34 CFR 685.102(b). The Department's portfolio reporting uses 361 days instead, for consistency with older FFEL reporting, and says so in a note to its own file. On 31 March 2026 there were 520,000 recipients between the two thresholds.

How many people are in default on student loans? The published figure of 9.0 million is a loan-level count of recipients, not a headcount of people, and on a parent PLUS loan the Department defines the recipient as the child rather than the borrowing parent. The status counts across the portfolio sum to 45.5 million against 42.6 million unduplicated recipients, so the double counting is measurable.

Are parent PLUS loans included in the cohort default rate? No. The statute names Stafford, SLS, Direct Subsidized and Direct Unsubsidized loans and the consolidation portion that repaid them. Grad PLUS and parent PLUS, $248.2bn between them, are outside the measure.

What happens when a student loan defaults? The debt moves into collection. Up to 15 percent of disposable pay may be garnished without a court judgment after 30 days of written notice. A defaulted Direct Loan can be rehabilitated once, through nine voluntary payments made within 20 days of their due dates across ten consecutive months.

Where does this data come from? The cohort rates come from the Department of Education's national default rate briefings and the school-level file published with them. The portfolio counts come from Federal Student Aid's quarterly Portfolio by Loan Status and delinquency reports. The balance measures come from the New York Fed's Quarterly Report on Household Debt and Credit. The two tables compiled for this piece are linked above.