Federal Reserve Payments Study — 2015–2024
The American payment system just posted its biggest decade of growth on record. Almost none of that growth happened where executives are looking.
Ten years of Federal Reserve data, read against the grain — six shifts in how American money actually moves, and what each one asks of the people who run banks.
The Argument
Read the Federal Reserve's latest Payments Study quickly and you get the familiar story: cards keep winning, checks keep dying, digital keeps rising. Read it closely and a stranger picture appears. The system Americans touch — cards, taps, wallets — is four out of every five transactions but barely one dollar in twelve. The system that actually moves the money — ACH and the large-value rails behind it — is nearly invisible to the people using it, and carries three-quarters of the value.
Most banks are organized, branded, and defended around the layer customers see. The economics, the risk, and the disruption are accumulating in the layer they don't. What follows traces six places where the two layers have pulled apart since 2015 — and argues that the surface calm of "cards up, checks down" is hiding a quieter, more consequential rewiring underneath.
Every figure here is the Fed's own, verified against the underlying workbooks. The interpretation is ours.
01 — The Great Divergence
In 2024, Americans made 187.7 billion card payments — 79% of all noncash transactions — worth $11.5 trillion. In the same year they made 39.7 billion ACH payments — 17% of transactions — worth $104 trillion. Cards win the count by more than four to one. ACH wins the value by nearly ten to one.
The temptation is to file this under quirks of averages. It isn't. It's a map of two distinct franchises that happen to share a P&L. One is a high-frequency, low-value, experience-and-loyalty business measured in taps and basis points of interchange. The other is a low-frequency, high-value, trust-and-plumbing business measured in float, settlement risk, and deposit balances. They reward different capabilities, fail in different ways, and are attacked by different challengers.
When a fintech says it is "disrupting payments," ask which system it means. Nearly all consumer fintech competes for the 8% of value that rides on cards. The 74% that rides on ACH is contested by a different, quieter set of players — including your own corporate-treasury clients.
Toggle between the count of transactions and the value they carry. The same four instruments; two opposite pictures.
Source: Federal Reserve Payments Study, 2024 (IDR data).
Why it matters If your payments strategy is one strategy, you are almost certainly over-invested in the layer customers see and under-invested in the layer that holds the money.
02 — Credit's Quiet Coup
Credit transactions (bars) re-accelerated to 9.6% a year in 2021–24 while debit growth (line) fell to its weakest on record — the first time credit has outgrown debit since 2000.
Source: Federal Reserve Payments Study, 2015–2024.
From 2021 to 2024, credit-card transactions grew 9.6% a year while debit grew just 4.1% — its weakest showing in any period the Fed has measured. The comfortable reading is that the American consumer is confident and issuers are winning. The uncomfortable reading is that three things are happening at once, and only one of them is good news.
First, affluent households are optimizing: rewards cards are rational when interchange funds the points. Second, the rails tilted the field — credit carries roughly 2.35% interchange plus lending margin; debit's interchange is capped and thin. The industry didn't just win on credit; it was paid to prefer it. Third, and least comfortable: some of this is borrowing, not spending. Credit-card balances crossed $1.28 trillion in late 2025, up 5.5% on the year, with delinquencies ticking up.
The rewards that drive credit growth are not free. The Fed's own research puts the transfer at $15.1 billion a year — flowing, on net, from less-affluent to more-affluent cardholders. One of the largest and least-discussed redistributions in consumer finance sits inside your most celebrated product.
Why it matters A franchise growing on interchange-funded rewards at the top of a credit cycle carries two risks at once — regulatory and cyclical. Enjoy the growth; underwrite the reversal.
03 — The Atomization of Money
Look at what a single transaction is worth, by instrument, and the system pulls to two poles. On one side, cards: an average ticket of $61, essentially flat for a decade — and debit's actually fell, from $43 to $41, since 2021. On the other side, checks: the average check is now worth $2,653, up 73% since 2015, with ACH beside it at $2,622. Cash tells the same story from the ATM: fewer trips, but $210 a withdrawal, up 57%.
This is what an instrument looks like when it specializes on the way to the exit. Checks didn't simply decline; they concentrated. Every low-value check that could migrate to a card or an app already has. What remains is the high-value, workflow-bound B2B and settlement core — the checks hardest to kill precisely because they carry the most money and sit deepest in accounts-payable habits. Meanwhile the low end fragments further: BNPL originations hit $45 billion at an average loan of $135; peer-to-peer transfers were the fastest-growing category in the entire study.
"Eliminate checks" programs are sold as cost-out. But the checks still standing are the most valuable, most fraud-exposed, and most operationally entrenched of all. The last tenth is where the real money and the real risk live.
A log scale flattens the poles so the divergence is visible: the high-value rails climb, cards stay pinned to the floor, and nothing settles in the middle.
Source: Federal Reserve Payments Study, 2015–2024; CFPB (BNPL).
Why it matters Per-transaction economics reward whoever owns the fragmenting low end; float and balance economics reward whoever owns the concentrating high end. A pricing model built for the old middle fits neither.
04 — Cash's Great Decompression
Transactional use (left axis) has halved since 2019. Precautionary holdings (right axis) spiked in the pandemic and settled above pre-2020 levels. Opposite trajectories.
Source: Fed Diary of Consumer Payment Choice, 2019–2024.
Transactional cash has genuinely collapsed: from 26% of consumer payments in 2019 to 14% in 2024, and for the first time in the Fed's diary history, cash is no longer even the most-used instrument for payments under $25. If that were the whole story, the "war on cash" would be over.
It isn't the whole story. The cash Americans hold — as a store of value, a backup, a hedge — spiked in the pandemic and has settled above its pre-2020 level, near $306 a person. Ninety-four percent of consumers still touched cash in the past month; two-thirds of the cash payments that remain are made by people who preferred another method and used cash anyway. Cash has stopped being a way to transact and become a way to feel safe.
A medium of exchange and a store of value have opposite futures inside the same banknote. One is trending to marginal; the other has found a floor. Treating "cash" as a single declining line is the most common analytical error in payments strategy.
Why it matters Branch, ATM, and cash-logistics decisions modeled on a straight line to zero will overshoot. Cash is now a resilience and inclusion asset, not a transaction volume — plan it as one.
05 — The Fraud Migration
The chip card worked. After EMV, fraud at the physical point of sale fell hard. But fraud didn't fall — it moved. By 2016, card-not-present fraud had overtaken in-person fraud to become the majority of card losses, running at nearly twice the rate per dollar. It moved online at the exact moment commerce did: card-not-present climbed from 21% of card volume in 2015 to 36% by 2022, and crossed half of all card value around 2020.
The lesson is uncomfortable for anyone planning the next control. Security investment doesn't destroy fraud; it reprices and redirects it toward the channel of least resistance — reliably, the newest and fastest-growing one. The United States is the standing proof: 26% of the world's card volume, but 42% of the world's card-fraud losses, concentrated in the remote channel the rest of the system is racing toward.
The next "remote" channel is already here: payments initiated by AI agents on a customer's behalf. It has no settled liability model, no chargeback framework, and every structural feature of the card-not-present channel that fraud has favored for a decade.
As EMV cut in-person fraud, the card-not-present rate rose above it (bars) — while card-not-present's share of volume climbed toward the majority (line).
Source: FRPS payments-fraud data (latest granular, 2016); NPIPS volume shares.
Why it matters If fraud follows commerce into whatever channel is newest, remote-first risk is not a line item — it is the baseline. Size controls, reserves, and liability terms as if every payment is card-not-present. Increasingly, it is.
06 — The Disruption Is Not Where You're Looking
Every US instant rail, set beside the ACH network they were meant to disrupt. The growth rates are explosive because the bases are tiny.
Source: frbservices.org, The Clearing House, Early Warning, Nacha (2025).
Instant payments are real and rising — and still a rounding error. In 2025, FedNow moved $853 billion, RTP cleared well over a trillion, Zelle crossed $1.2 trillion. Impressive, until you set them beside the incumbent they were built to disrupt: the ACH network moved $93 trillion. Every US instant rail combined is a low-single-digit fraction of it. The brake is structural and domestic — Brazil put Pix in the hands of 76% of its population in under five years, and India's UPI clears more in a day than US real-time rails clear in a year, because both were run as public utilities. America's are voluntary, and the banks opt in one at a time.
Stablecoins are the sharper illusion. You have seen the headline: roughly $33 trillion moved on-chain in 2025, "more than Visa and Mastercard combined." Strip the bots, wash trades, and automated churn and genuine activity is nearer $28 trillion. Filter to actual payments — real goods, real invoices — and it is about $390 billion. Roughly one percent of the headline.
Cite the definition, not the number. An executive who repeats "stablecoins beat Visa" has misread the threat by a factor of a hundred — and will defend the wrong flank. The exposure is cross-border B2B and FX margin, not the checkout page.
The same year, three numbers, two orders of magnitude apart. What you measure decides which threat you prepare for.
Source: on-chain aggregators; Chainalysis (adjusted); McKinsey–Artemis (payments-only). Estimates.
The real $390 billion is doubling yearly and lives almost entirely in cross-border B2B — the correspondent-banking and FX-margin business, not domestic checkout. And the genuinely under-noticed shift sits behind all of it: with a US central-bank digital currency legally off the table through 2030, the contest that decides whether deposits stay on bank balance sheets is tokenized deposits versus stablecoins. The GENIUS Act's quiet ban on interest-bearing stablecoins may prove the most important piece of deposit protection the industry never asked for.
Why it matters Point your defenses where the money moves: interchange erosion at the small-ticket high-frequency end, correspondent-banking disruption at the cross-border end, and deposit tokenization as the battle for the balance sheet itself.
The Agenda
You almost certainly report card volume weekly and ACH value quarterly. Reverse it. The money — and the risk — lives where the dollars are.
Model credit growth with interchange contested and the credit cycle turning. Record 2022–24 profit was partly a rate-cycle windfall; don't capitalize a windfall as a franchise.
Build economics for a fragmenting low end (per-transaction) and a concentrating high end (float and balances) — not for a middle that no longer exists.
Retire the straight-line-to-zero model. Size ATM, branch, and cash logistics for a resilient, inclusion-critical store of value.
Fraud follows commerce online, and agentic payments arrive without a liability model. Set reserves and controls to a card-not-present baseline.
The decisive tokenization question is whether deposits stay on your books. Engage tokenized deposits now; treat stablecoins as a cross-border B2B threat.
Methodology & Sources
The spine of this report is the Federal Reserve Payments Study (2015–2024 release), verified against the underlying IDR, NPIPS, and payments-fraud workbooks rather than the summary tables alone. Card-not-present and alternative-payment detail are published through 2022; the latest granular FRPS fraud data is 2016 and is labeled as such throughout. Instant-rail, interchange, consumer-diary, household-credit, BNPL, and stablecoin figures are drawn from the Federal Reserve System, Nacha, The Clearing House, Early Warning Services, the CFPB, the New York Fed, and the Nilson Report; third-party estimates are flagged where used.
Figure-level citations are in the repository's SOURCES.md. The data is the Federal Reserve's; the interpretation is the author's own and carries no affiliation with, or endorsement by, the Federal Reserve.