Student Loan Delinquency: 7.83% vs 10.6%

On August eleventh, the New York Fed published its quarterly household debt report. Coverage of student loans treated it as almost entirely good news: the share of student-loan balances newly falling into serious delinquency came in at seven point eight three percent, down from twelve point eight eight percent a year earlier. A five-point drop.

The same report says the share of student-loan balances actually sitting ninety or more days past due rose to ten point six percent, up from ten point three percent in the previous quarter.

Both numbers are in the same document, and neither is wrong. On the morning of the twelfth, I read the coverage and opened my own file. A ninety-day mark from 2025 was still there, exactly where it had been a month earlier. The missing distinction is that the two figures measure different things.

The report covers the second quarter of 2026. Total household debt fell by thirteen billion dollars, about one tenth of one percent, to eighteen point eight trillion. Student-loan balances fell by seven billion to one point six five trillion. Mortgage balances fell by seventy-four billion, though the Fed says that drop mostly reflects a servicer-transfer gap in mortgage reporting; otherwise, balances would have stayed flat. Those were the only two reported categories to fall. Across all debt types, four point seven percent of outstanding balances were in some stage of delinquency.

Then the two student-loan delinquency figures, which is where the coverage went wrong.

The first is a transition rate: the share of balances that crossed into serious delinquency, ninety or more days past due. The Fed publishes it as a four-quarter moving sum, so it covers the year ending in June rather than the three months of the quarter on their own. That came in at seven point eight three percent, against twelve point eight eight percent a year earlier. It counts arrivals over a rolling window.

The second is the share of student-loan balances currently ninety or more days past due, which rose to ten point six percent from ten point three percent in the first quarter. It counts every dollar of balance sitting in that state, whenever it arrived.

The first fell year over year. The second rose quarter over quarter. Almost every piece of coverage I read reported the first and none mentioned the second.

The Fed adds a caveat of its own that went unquoted in everything I read: student loan delinquencies were an exception this quarter, and the continued re-reporting of defaulted student debt is causing some distortions in the figures. A number its own publisher flags as distorted is a strange thing to build a headline on.

This is the whole article in two words.

A flow measure counts movement over a period: how many balances went seriously delinquent across the span it measures. A stock counts what exists at a point in time: how much is currently sitting in that condition. The Fed's transition rate is a flow. The Fed's ninety-plus share is a stock, counted in balances. Your credit report is a stock too, counted in entries.

When a flow rate falls, it means fewer balances are entering the state. It does not mean any have left it. Mine went ninety days late in the early months of 2025, which sits outside the year this rate covers, so it is not in the Q2 2026 figure. I was counted in the window that contained me, and I will not be counted again. From the perspective of this statistic I have already happened.

This is why a falling national number can feel at odds with your own file. A slower inflow does not reduce the balances already in delinquency, and it does not explain why that total rose. Stocks change through new delinquencies, cures, payoffs, and reporting changes at the same time. The Fed warned that re-reporting defaulted student debt was distorting these figures. What a slower inflow does not do is remove an existing mark. For an accurate late payment, that mark follows the seven-year reporting clock.

And the weight of it fades, which is different from removal. A ninety-day mark from 2025 counts for less in 2028 than it did in 2025, even though it is still on the file. That fading is real and it is how a clean stretch gradually gains ground, but it is not visible as an event and it is not what the Fed measured.

The two require opposite responses.

• If your mark is accurate and you were genuinely late, the clock is the answer and there is no shortcut.
• If your mark arrived after a servicer transfer, dispute it with whichever servicer actually reported it and with each bureau showing it.
• If a single loan appears twice, that is a duplicate reporting error and it is disputable.
• If the status contradicts the repayment plan you were actually on, gather the plan documentation before you file anything.
• If you are current now and the mark is historical, the useful work is adding positive history rather than fighting the past.

The distinction matters because disputing an accurate mark wastes months and achieves nothing, while accepting an inaccurate one costs you years of a report you did not earn. Given that reporting in this population was questioned at the congressional level, checking rather than assuming seems like the reasonable default.

Total household debt fell slightly and two large categories shrank, which is a different picture from the one most people carry around. Four point seven percent of all outstanding balances were in some stage of delinquency.

I cannot tell you what this means for your loan application. A quarterly report cannot reveal the connection between aggregate delinquency trends and individual underwriting decisions. Lenders set their own criteria and do not publish how macro conditions affect them.

What I will say is that "the national numbers are falling" is not an argument available to you in an application. The decision gets made on your file. The aggregate is context for economists, not a fact about you.

I pulled all three reports and checked the student-loan entries line by line against my own payment records. I was looking for three specific things: whether the same loan appeared more than once, whether the date of first delinquency matched what actually happened, and whether the reported status matched the repayment arrangement I had been on at the time.

Mine was accurate. The mark is real, I earned it, and it will come off on its own schedule. That was not the answer I was hoping for but it was worth the hour, because the alternative is how an error survives for seven years: assume it is correct and never check.

The check is simple: those three questions are the whole task. A mark on your report is a statement from a furnisher about you. Furnishers have accuracy obligations and disputes have an investigation process, but neither means that a particular line has already been checked against your records. Only that comparison can settle it.

Quarterly debt statistics generate headlines every three months, and the headline is almost always a flow measure presented as if it described a stock. "Delinquencies fall" reads as "fewer people have delinquencies." What it usually means is "fewer balances crossed the line during whatever period the statistic covers."

The test is simple. Ask whether the number counts events during a period or conditions at a point in time. Transition rates, new originations, and quarterly changes are flows. Total balances outstanding, the share of accounts currently delinquent, and what is on your report are stocks. A falling flow signals a trend, not how much has accumulated. Your credit file is an accumulated position.

Five points of movement in one number, three tenths of a point of deterioration in another, and one unchanged line on my report. None of the three is wrong.

If you are carrying a student-loan delinquency from the return-to-repayment period, the useful actions are limited and none involve the Fed's numbers. Check whether the mark is accurate, because reporting in this population drew a congressional letter. If it is not, dispute it with the servicer that reported it and with each bureau showing it. If it is accurate, stop looking at it and focus on the next twelve months of clean history.

And when the next quarterly report comes out, check what the measure counts before the number. A falling transition rate is genuinely good news about the future, but it says nothing about the past already on your file.

1. Did student loan delinquencies rise or fall in 2026?

Both, depending which figure you mean. The New York Fed reported on August 11, 2026 that the transition rate into serious 90-plus-day delinquency, published as a four-quarter moving sum, was 7.83% in Q2 2026, down from 12.88% a year earlier. Yet the share of balances actually 90+ days past due rose to 10.6% from 10.3% in Q1. The first counts new arrivals; the second counts everyone already there.

2. Why did my credit report not move when national delinquencies fell?

Because the two measure different things. The Fed reported a transition rate, which counts arrivals into delinquency over a rolling four-quarter window. Your report records a delinquency you already have. A falling inflow has no mechanism to remove an existing mark.

3. What is the difference between a flow and a stock in credit data?

A flow counts events during a period, such as balances transitioning into delinquency over the period the measure covers. A stock counts conditions at a moment, such as what is currently on your report. Quarterly headlines are usually flows presented as though they described stocks.

4. How do I remove a student loan delinquency from my report?

If it is accurate, time is the only answer: a late payment is generally reportable for about seven years from the delinquency, and paying the loan current or off does not delete the historical mark. If it is inaccurate, dispute it with the servicer that reported it and with each bureau showing it. An investigation has to happen, but correction or deletion follows only if the information turns out to be inaccurate, incomplete, or unverifiable.

5. Were student loan credit reporting errors common after return to repayment?

The period produced a documented pattern of problems including duplicate loan records after servicer transfers and statuses inconsistent with borrowers' actual repayment plans, and it drew a 2026 letter from senators to all three nationwide bureaus. That is a reason to check your own entries rather than assume they are right.

6. How much household debt do Americans hold?

The New York Fed put total household debt at $18.8 trillion at the end of Q2 2026, down $13 billion on the quarter, with student-loan balances at $1.65 trillion and 4.7% of all outstanding balances in some stage of delinquency.

7. Does a falling national delinquency trend help my loan application?

Not as an argument you can make. A lender reads your file, not the aggregate, and the Fed report does not show whether or how any particular lender or automated system uses aggregate trends. The practical response is to work on the file rather than to cite the trend.