Blog
Five ways credit card delinquency headlines can mislead you
One quarter of card data can read as better and worse at once. Five reading traps, with a worked example and the Q3 2026 New York Fed and TransUnion figures.
In November 2026 two respected sources published third-quarter card data within a week of each other, and a careful reader could take away opposite messages. The New York Fed said the flow of card balances into serious delinquency was slightly better than a year earlier. TransUnion said one of its delinquency measures was at its lowest level in three years while another was higher than a year earlier. Neither source made a mistake. Each measured something different, and a headline that compresses them into "card delinquencies are rising" or "falling" throws away the part that tells you what is actually happening.
This post walks through five ways a delinquency headline can mislead, using the real Q3 2026 figures where they are verified and a labeled hypothetical where a made-up example explains the mechanism better. None of it is a forecast. The aim is to make you a slower, more skeptical reader of the next number you see.
First, what "delinquent" means in these reports
A card account is in delinquency when a required payment has not arrived by its due date. Credit reporting and industry data group delinquency by how late the payment is: 30 days, 60 days, 90 days and beyond. Reports about the health of the card market usually focus on serious delinquency, which is the term for being 90 or more days past due, because those accounts are unlikely to catch up without help and are heading toward a charge-off, the issuer's accounting write-off after a long period of missed payments. A charge-off is not forgiveness. The debt is still owed.
Early delinquency, say one missed payment that is fixed within the month, is far more common and far less meaningful for the market as a whole. When a headline says "delinquencies," check which bucket it means. Most of the figures below are the 90-plus-day kind.
Trap one: a flow rate is not a stock rate
The New York Fed's quarterly Household Debt and Credit report published on November 10, 2026 said that the flow of credit card balances into serious (90-plus-day) delinquency was 6.87 percent, annualized, in the third quarter of 2026, compared with 7.05 percent in the third quarter of 2025. The New York Fed described that as slightly improved.
That is a flow measure. It asks how much of the balance that was current or in early delinquency at the start of the quarter newly crossed into serious delinquency during it, and then expresses the result as an annual rate. A stock measure asks something different: of all balances outstanding today, what share is seriously delinquent right now? The New York Fed page read for this research did not give a stock rate for cards, and this post does not quote one.
The two behave differently. A stock includes every balance that crossed the line in earlier quarters and has not yet been paid, cured or written off. A flow counts only the new arrivals.
Here is a hypothetical to make the difference concrete. These numbers are invented for illustration and are not the market's.
- Say a lender has $1,000,000 in card balances at the start of a quarter.
- During the quarter, $15,000 of that moves from current into 90-plus-day delinquency. That is 1.5 percent of $1,000,000 in a single quarter.
- Annualizing in the usual way, by multiplying the quarterly share by about four, gives roughly 6 percent. (The exact method in a published series is the publisher's; check its notes.)
- Meanwhile, suppose $40,000 of the lender's balances are sitting at 90-plus days today, including accounts that crossed in earlier quarters. That stock is 4 percent of $1,000,000.
A reader who sees "6 percent" in one article and "4 percent" in another may think the sources disagree. They do not: one number is a rate of arrival over a year and the other is a snapshot of who is in the room. If the arrival rate drops but the room is still crowded, the stock can stay high even while the flow headline reads "improved." The reverse is also possible. The practical habit is to ask of any delinquency figure: is this new delinquencies, or all delinquencies?
Trap two: balances and borrowers are different denominators
TransUnion's Q3 2026 Credit Industry Insights report, released on November 5, 2026, gave two 90-plus-day delinquency figures for bankcards. At the balance level, delinquency was 1.99 percent in Q3 2026 against 2.05 percent a year earlier, which TransUnion called the lowest level since the third quarter of 2023. At the borrower level, it was 2.50 percent against 2.37 percent. The two measures moved in opposite directions in the same quarter.
How can that happen? A balance-level rate divides dollars that are seriously delinquent by all dollars outstanding. A borrower-level rate divides people who are seriously delinquent by all people with a card balance. If the delinquent group has smaller balances than the average borrower, or if the dollars owed by borrowers in good standing grow faster than the dollars owed by borrowers in trouble, the two rates can diverge.
A hypothetical shows it. Take a group of 100 cardholders, with invented balances.
Year one. Two borrowers are 90-plus days delinquent and each owes $4,000, so $8,000 is delinquent. The other 98 each owe $2,000, which is $196,000. Total balances are $204,000.
- Borrower-level rate: 2 of 100 = 2.0 percent.
- Balance-level rate: $8,000 / $204,000 = 3.92 percent.
Year two. Three borrowers are delinquent and each owes $2,500, so $7,500 is delinquent. The other 97 each owe $2,100, which is $203,700. Total balances are $211,200.
- Borrower-level rate: 3 of 100 = 3.0 percent.
- Balance-level rate: $7,500 / $211,200 = 3.55 percent.
The borrower-level rate went up by a full point. The balance-level rate went down by about 0.4 of a point. Both statements are true of the same group, and both are accurate headlines. One tells you that more households are in trouble. The other tells you that the dollars at risk shrank relative to the total. Which matters depends on who is asking: a lender worried about losses cares about the dollars, while a counselor or a policymaker worried about households cares about the people.
This example is not meant to explain why TransUnion's two numbers moved the way they did. The report's figures are real, but the reasons are not in the verified material behind this post. The example shows only that opposite moves are not a contradiction.
Trap three: "household debt" is not "card debt"
A third source of confusion is the scope of the debt being measured. The same New York Fed report said that total U.S. household debt was $19.105 trillion at the end of the third quarter of 2026, and that credit card balances were $1.285 trillion of it. It also said that 5.0 percent of all outstanding household debt was in some stage of delinquency, up 0.3 percentage points from the second quarter, and that the flow of all debt into serious delinquency was 2.64 percent against 3.03 percent a year earlier.
Those lines describe household debt as a whole: mortgages, auto loans, student loans, cards and other balances together. A headline that says "5 percent of household debt is delinquent" is not a statement about credit cards. And the all-debt flow rate (2.64 percent) is lower than the card flow rate (6.87 percent) in the same release, which is a reminder that cards behave differently from the rest of the mix. Card balances are revolving credit, unsecured, with no collateral behind them and no fixed end date, and a mortgage is neither of those things.
Likewise, a figure that mixes the New York Fed's $1.285 trillion in card balances with TransUnion's $1.16 trillion in bankcard balances for the same quarter is comparing two different counts of the same market. The two sources do not draw the box in the same place, a point taken up in card industry numbers explained: purchase volume, balances and accounts. A delinquency percentage is only as meaningful as the balance it sits on top of.
Trap four: the average hides who is affected
The New York Fed also published an analysis on November 10, 2026 on its Liberty Street Economics blog, looking at how card delinquency differs by income. According to that analysis, card delinquency rose sharply from 2022 to 2024, and it rose most in lower-income ZIP codes. The same analysis reported that the median credit score at card issuance fell from 699 (2017 to 2019) to 688 (2020 to 2022) and has since recovered to a level above where it stood before the pandemic. The authors said they expect new card delinquencies to keep improving, unless there is a macroeconomic shock.
Two things follow. First, a national rate is an average over very different households. If the increase is concentrated in lower-income areas, then a national rate that moves a little can hide a large change for a specific group, and a national rate that improves can leave that group no better off. Second, the age and quality of the accounts matter. A group of cards opened when approvals were looser behaves differently from a group opened under tighter standards, and the headline rate blends them.
The Federal Reserve's 2025 household survey gives a hint of the unevenness on the borrowing side. Among cardholders, 52 percent of those with incomes under $25,000 carried a balance at least once in the prior 12 months, 57 percent in the $25,000 to $49,999 band, 50 percent in the $50,000 to $99,999 band and 37 percent at $100,000 or more. Carrying a balance is not delinquency, and it is not even a problem in itself. But where balances are carried, the room for a missed payment is larger.
For your own reading, the lesson is plain. A national average is a statement about the market. It says little about your household, your neighborhood or your card.
Trap five: direction depends on the baseline and the window
The last trap is the simplest. "Better" and "worse" depend on what you compare to. The 6.87 percent card flow rate is lower than the 7.05 percent of a year earlier, which is the comparison the New York Fed made. The release's all-debt delinquency line, at 5.0 percent, was compared with the previous quarter and rose by 0.3 points. One release, two windows, two directions.
Three habits help here.
- Check the window. Is the comparison to last quarter, to a year ago, or to some earlier peak? Quarter-to-quarter comparisons mix in seasonal patterns that year-over-year ones remove.
- Check the base. Improving from a high level is still a high level. The New York Fed's own analysis says card delinquency rose sharply from 2022 to 2024, so "slightly improved" is a statement about direction, not about whether the level is comfortable.
- Check the size of the move. A fall from 7.05 to 6.87 is 0.18 of a percentage point. TransUnion's charge-offs, as reported secondhand, were about $15.76 billion in Q3 2026 against $15.79 billion a year earlier, a difference of about $30 million on a base of roughly $15.8 billion. That is a tiny move, and a headline that says "charge-offs fall" is technically right and practically empty.
The Q3 2026 figures side by side
| Measure | Source | Q3 2026 | A year earlier | What it measures |
|---|---|---|---|---|
| Flow of card balances into 90+ day delinquency, annualized | New York Fed | 6.87 percent | 7.05 percent | New arrivals into serious delinquency |
| Flow of all debt into serious delinquency | New York Fed | 2.64 percent | 3.03 percent | New arrivals, all kinds of debt |
| Share of all household debt in some stage of delinquency | New York Fed | 5.0 percent | Up 0.3 point from Q2 2026 | Stock, all debt |
| Bankcard 90+ day delinquency, balance level | TransUnion | 1.99 percent | 2.05 percent | Dollars delinquent / dollars owed |
| Bankcard 90+ day delinquency, borrower level | TransUnion | 2.50 percent | 2.37 percent | People delinquent / people with a balance |
One row is not like the others: the 5.0 percent line is compared with the prior quarter, not with the same quarter last year. Putting the numbers in one table makes that visible. That is the whole technique: put the definition next to the number before drawing a conclusion.
A short checklist for the next delinquency headline
Run each of these in thirty seconds.
- Which debt? Cards, bankcards, or all household debt?
- Which bucket? 30 days, 60 days, 90 days or more? Or "some stage of delinquency"?
- Flow or stock? New arrivals during a period, or everything outstanding at a point in time?
- Dollars or people? Balance-level or borrower-level?
- Whose data? The New York Fed, TransUnion and the Federal Reserve each count the market their own way, so their totals and rates are not interchangeable.
- Compared with when? Last quarter, last year or a peak?
- How big is the move? A tenth of a point can be the entire story of a headline.
If the article does not tell you the answers, that is information too. Look for the source's own page. The links in the New York Fed and TransUnion releases tell you more than a summary can.
What this means for your own household
A market delinquency rate does not predict whether you will have trouble paying. What it does tell you is how the market around you is behaving, and that can show up in your life in two ways. Lenders may tighten or loosen approvals and limits depending on what they see in losses, and the terms on offer to you may change with them. And if you are carrying a balance, the national figures are a reminder that many households do, and that falling behind is common enough that the issuer has processes for it.
Your own signals matter more. Am I in too much credit card debt? Seven signs to check yourself lists the signs that a balance is becoming a problem long before it reaches 90 days, and what national credit card debt figures mean for your own household explains how to translate a national total into a question about your own statement. If you have already missed a payment, call the number on the back of your card and ask what options exist before the account ages further. If a debt collector or an issuer is treating you unfairly, you can file a complaint with the Consumer Financial Protection Bureau. Late payments can stay on a credit report for up to seven years, which is why the early weeks matter.
Key takeaways
- A delinquency flow rate counts new arrivals into 90-plus-day delinquency over a period. A stock rate counts everyone who is there at a point in time. They answer different questions and should not be compared.
- Balance-level and borrower-level rates can move in opposite directions, as TransUnion's Q3 2026 figures did (1.99 percent and 2.50 percent), because the dollars and the people are not the same group.
- "Household debt delinquency" covers mortgages, autos, student loans and cards together. The New York Fed's card flow rate (6.87 percent) and its all-debt flow rate (2.64 percent) in the same report show how different they are.
- A national average hides who is affected: the New York Fed's analysis says the rise from 2022 to 2024 was greatest in lower-income ZIP codes.
- Before trusting a direction word like "improved," check the comparison window, the starting level and the size of the move.