Blog

Card industry numbers explained: purchase volume, balances and accounts

Purchase volume, balances, accounts and limits measure different things. Why Nilson, the New York Fed, the Fed and TransUnion disagree, with 2025-2026 figures.

Search for the size of the U.S. credit card market and you will find a trillion-dollar figure, a different trillion-dollar figure, a count of hundreds of millions of cards and a per-person average that matches none of them. They are not wrong. They are answers to four different questions: how much people spent, how much they owe, how many accounts exist and how much they are allowed to borrow. A headline that swaps one for another can sound alarming or reassuring for no reason other than the swap.

This explainer defines each measure, shows which dated figures are verified for 2025 and 2026 and gives you a short method for sanity-checking any statistic about the card business before you repeat it. It also flags the figures that circulate widely but could not be verified, so you know which ones to leave alone.

The four questions behind every card statistic

Think of a card market as a river with a reservoir. Money flows through it every day, some of it stays behind and the reservoir has a size limit.

  • Purchase volume is the flow: the dollars spent on cards over a period, usually a year. It is mostly paid off within weeks and does not stay on anyone's books.
  • Balances are the reservoir: the dollars owed at a point in time, including purchases that have not been paid yet and money that is being carried from month to month.
  • Accounts are the number of cards or credit lines that exist, whether or not they carry a balance or are used.
  • Credit limits are the reservoir's capacity: the total the issuers have agreed to lend. Most of it is never used.

Each is reported by different organizations, using different data, over different periods. Treat a number without its question attached as unfinished.

Purchase volume: what Nilson counts

The Nilson Report tracks card purchase volume by card network. For 2025 it reported that Visa, Mastercard, American Express and Discover cards issued in the United States generated $11.463 trillion in purchase volume across all products, up 6.4 percent from 2024. Credit cards on those four networks accounted for $6.512 trillion of that, up 6.1 percent. The remainder is mostly debit and prepaid cards.

By network, Nilson reported Visa-branded purchase volume of $7.028 trillion (all products, up 6.8 percent), with Visa credit at $3.412 trillion, and Mastercard-branded purchase volume of $2.958 trillion (up 6.3 percent). Visa's share of the combined Visa-plus-Mastercard total of $9.986 trillion was 70.38 percent. A useful check: $7.028 trillion plus $2.958 trillion is $9.986 trillion, and $7.028 trillion divided by $9.986 trillion is about 0.7038.

You can do a little more arithmetic with these totals, but you should say that it is arithmetic and not Nilson's own claim. If $11.463 trillion is the all-products total, then Visa is about 61 percent of it ($7.028 trillion over $11.463 trillion), Mastercard about 26 percent and everything else (mostly American Express and Discover) about 13 percent, the residual of $1.477 trillion. These are derived shares. Issuer-by-issuer market shares are not given here because the detailed tables are behind a paywall and could not be verified, so claims such as "this bank has X percent of the market" should not be repeated from this post. For what the network itself does in a transaction, see what happens in the seconds after you tap: network, issuer and merchant and the profiles for Visa and Mastercard.

Two more things about purchase volume. It counts what was spent, not what was borrowed, so a household that pays in full every month adds to the total without ever adding to the balance. Volume grows with prices as well as with behavior, which is one reason why card spending totals can rise even when people buy no more is worth reading before treating a record total as a sign of anything.

Balances: three sources, three totals

Now the reservoir. Three respected sources reported card or revolving balances for the third quarter of 2026, and they disagree.

Source Figure Date What it covers
New York Fed, Household Debt and Credit $1.285 trillion End of Q3 2026 Credit card balances
TransUnion, Credit Industry Insights $1.16 trillion Q3 2026 Total bankcard balances
Federal Reserve, G.19 consumer credit $1,368.5 billion, seasonally adjusted September 2026 Revolving credit, mostly cards

The New York Fed said balances rose $22 billion from the second quarter and $52 billion from the third quarter of 2025. TransUnion's year-earlier figure was $1.11 trillion. The gap between the first two levels is about $125 billion ($1.285 trillion minus $1.16 trillion), and the Fed's measure is higher still. Yet the growth is similar: the New York Fed's increase of $52 billion on a base of about $1.233 trillion is 4.2 percent, and TransUnion's rise from $1.11 trillion to $1.16 trillion is about 4.5 percent. Those percentages use rounded totals, so they are approximate.

Why do the levels differ? Revolving credit, the Fed's category, includes more than the cards the New York Fed counts, so those two series should not be expected to match. (revolving credit is the general term for borrowing you can repeat up to a limit and repay over time.) The reasons for the gap between the New York Fed and TransUnion totals are not set out in the material behind this post, and it is more honest to say so than to guess. The practical lessons are:

  • Use one source consistently when you track a change over time. Compare a series against its own past, not against another source's total.
  • When a headline gives a level, ask whose level it is. When it gives a change, ask whether it is the same series.
  • A seasonally adjusted figure and a not-seasonally-adjusted figure from the same source differ. The G.19 revolving figure for September 2026 was $1,368.5 billion seasonally adjusted and about $1,323.2 billion not seasonally adjusted. Neither is wrong.

For how to read the Fed's release itself, see How to read the Federal Reserve monthly consumer credit report.

Accounts: cards are not people

TransUnion, one of the credit bureau companies, counted 590.5 million bankcards in Q3 2026, up from 567.5 million a year earlier, 545.1 million in 2024 and 530.6 million in 2023. That is a rise of 23.0 million over a year, or about 4.1 percent. TransUnion also said that 178.0 million consumers carried a balance on a bankcard and that 263 million consumers were credit-active across all credit types.

A count of accounts is not a count of people, and it is tempting to divide one by the other. Try it, and see why not to. Dividing 590.5 million bankcards by 263 million credit-active consumers gives about 2.2. That is not "cards per cardholder," because 263 million includes consumers with only a mortgage or a car loan and no bankcard, and because "bankcard" is TransUnion's own category. A per-person average like that is the kind of figure that circulates because the division is easy, not because it is right. The widely repeated claim of about 3.9 cards per adult could not be verified in any source, and Experian's figure of about 3.7 active cards per consumer comes from a single snippet with an unconfirmed base. Use "about 590 million bank credit cards, according to TransUnion in Q3 2026" and leave per-person claims out.

The Federal Reserve's 2025 household survey gives the person-level view. It found that 82 percent of U.S. adults had a credit card in 2025, down from a high of 84 percent in 2021. It also reported that 45 percent of cardholders carried a balance at least once in the prior twelve months, and 37 percent of all adults did. Those are shares of people, from a survey, on a different base from account counts.

Averages: what a balance per borrower really means

Average balances look simple and are the least comparable numbers of all. Three are verified for 2025 and 2026:

  • TransUnion reported average debt per borrower of $6,678 in Q3 2026, up from $6,523 a year earlier.
  • The Federal Reserve's household survey reported an average card balance of $7,279 among respondents with cards in 2025, up $748 (11 percent) from 2023 according to linked credit data.
  • Experian's 2025 Consumer Credit Review reported an average balance of $6,768 for 2025 against $6,730 in 2024 (secondary reporting).

All three are in the same neighborhood, and none is the same number, because each source chooses a different group, date and definition. Notice what each is not: none is "what a typical household owes," none says that the average household carries a balance and none says what you owe.

Here is a quick check that shows why the base matters. TransUnion's own totals give $1.16 trillion in balances across 590.5 million bankcards. Dividing gives about $1,964 per account. But that average includes every account, including the many with no balance. TransUnion's average per borrower, which counts only those carrying debt, is $6,678. Dividing $1.16 trillion by the 178.0 million consumers who carry a balance gives about $6,517, close to but not equal to $6,678. The difference shows that the published measures use definitions that a simple division of the rounded totals cannot reproduce. When a headline says "average balance," ask: average per account, per borrower or per cardholder?

Limits: capacity, not debt

The New York Fed reported that aggregate card credit limit totals rose $115 billion in Q3 2026. A rising limit is not a rising debt. It means issuers extended more room. The ratio of balances to limits is credit utilization, which is a scoring input for individuals, and an aggregate version of it can be computed only when you have the total limits, which are not among the verified figures here. Experian's 2025 review reported average card utilization of 29.1 percent (September 2025, flat versus 2024), while FICO data cited elsewhere put it at 35.5 percent. The two sources differ because of method, which is the same lesson again. No average credit limit is given, because no current named source could be verified, and any figure you see for "the average limit" deserves suspicion.

A household example: flow versus stock

Take a hypothetical household to see how a large volume and a small balance can both be true. These numbers are invented.

Say Jordan puts $3,000 of spending on a card every month and pays the full statement balance each month. Over a year Jordan generates $3,000 × 12 = $36,000 of purchase volume. On the day a statement closes, the balance might be about $3,000, and between payment and the next statement the balance may be near zero on average. If Jordan never pays interest, the card adds $36,000 to the industry's purchase volume and perhaps $3,000 to a balance snapshot. The ratio is 12 to 1.

Now take a second household, Kim, who spends $1,500 a month and carries a $6,000 balance from month to month. Kim adds $18,000 to purchase volume and $6,000 to balances, a ratio of 3 to 1. Kim contributes less to the volume headline and more to the balance headline, and nobody would say Kim is the bigger spender.

Apply the same idea to the verified totals. Nilson's 2025 credit-card purchase volume of $6.512 trillion divided by the New York Fed's $1.285 trillion of balances is about 5.07. That is a rough ratio of annual flow to a snapshot of stock from two different sources and two different periods, so it is only an order of magnitude. It does say that card balances are a small fraction of what runs through the system in a year, which is what you would expect when many cardholders pay in full. The point is not to compute a precise number. It is that volume and balance move for different reasons.

Which number answers which question

If you want to know ... use this kind of statistic, and say which source.

  • How much is spent on cards? Purchase volume, from Nilson (annual, by network).
  • How much do households owe? Balances, from the New York Fed or the Federal Reserve, and say which.
  • How many cards exist? Accounts, from TransUnion, and say it is bankcards.
  • How many people carry debt? Borrowers or survey shares, from TransUnion or the Fed's household survey.
  • How much room do they have? Limits, from the New York Fed's aggregate limit change.
  • How is credit quality? Delinquency measures. five ways credit card delinquency headlines can mislead you explains how those can be misread.
  • What does a household pay? Not an industry statistic: an APR figure and the household's own statement. Why every 'average credit card APR' you read is a different number explains why even the "average APR" has several answers.

A five-step check before you repeat a number

  1. Name the measure. Volume, balance, accounts or limits? If the writer does not say, do not use it.
  2. Name the source and the period. A figure without an "as of" date is a rumor.
  3. Name the base. Account, borrower, cardholder, adult or household?
  4. Match like with like. Compare a series to itself, not to another source. If you must compare, note the difference in definitions.
  5. Check the move against the size. A change of $22 billion on a base of $1.285 trillion is about 1.7 percent in a quarter (22 divided by 1,263). Say what it is.

On that last point the quarter's increase can be computed: the base at the start was about $1.263 trillion, and $22 billion over $1.263 trillion is about 1.7 percent. A number that sounds large in dollars can be modest in proportion.

What this means for your own spending

None of these totals tells you what to do with your cards. They describe a market. Your own statement is where to look for what matters: your balance, your limit, your utilization and your interest. what national credit card debt figures mean for your own household shows how to translate a national total into questions about your own account, and the household debt definition places cards among the other debts in the national total. Treat the industry numbers as background and your statement as the source.

Key takeaways

  • Purchase volume (spending), balances (debt), accounts (cards) and limits (capacity) answer four different questions. Credit-card purchase volume on the four networks was $6.512 trillion in 2025 per Nilson, while the New York Fed counted $1.285 trillion of card balances in Q3 2026.
  • Three respected sources gave three balance levels for Q3 2026: $1.285 trillion, $1.16 trillion and $1,368.5 billion (revolving, seasonally adjusted). Track change within one series, not across them.
  • TransUnion counted 590.5 million bankcards in Q3 2026. Do not divide that by the population: the 3.9-cards-per-adult claim could not be verified, and the base of other per-person figures is unclear.
  • Average balances depend on the base: about $1,964 per account, $6,678 per borrower in TransUnion's own measures. Ask what is being averaged.
  • Before repeating a statistic, name the measure, source, period and base, and check the move against the size.

Terms used in this article

Related issuers

Latest on this topic