Stablecoins in the money supply? The Fed just mapped how to count them — and merchants are the variable

What the Fed note actually says (September 4, 2026)

FEDS Notes are short research pieces in which Federal Reserve Board staff publish their own analysis — explicitly their views, not the Board's, and not part of any policy deliberation. That caveat matters and we will keep it in view. But the topics staff choose to formalize are a leading indicator of what the institution expects to need answers for, and on September 4, 2026, Kristen Payne and Mary-Frances Styczynski chose this one: how three new tokenized instruments — payment stablecoins, tokenized bank deposits and tokenized money market funds — map onto the US monetary aggregates.

The aggregates are the official measures of how much money exists. M1 covers the most liquid forms — currency and demand deposits, "readily accessible for spending." M2 adds less-liquid savings instruments: small time deposits, retail money market funds, assets whose components typically allow redemption within one to two business days. The monetary base counts currency plus reserve balances at the Fed. These series feed the Fed's H.6 release and, through it, decades of economic research and policy debate.

The note's findings, compressed: tokenized bank deposits are already inside the aggregates, because legally they remain ordinary deposits regardless of the ledger they move on. Retail tokenized money market funds are already inside M2, alongside their traditional siblings. Payment stablecoins are in neither M1, M2, nor the monetary base — an entire instrument class, sized at roughly $292 billion globally per contemporaneous coverage, invisible to the official count of US money. Against a US M2 of about $23.2 trillion, that is around 1.3% — small enough not to distort policy today, large enough that the staff thinks the measurement question can no longer be left unexamined.

Three tokenized assets, three different answers

The most useful thing in the note is not a conclusion but a sorting rule, applied consistently across the three instruments. Step one is functional: does the asset behave like a medium of exchange (highly liquid, immediately spendable, low-yielding → M1) or a store of value (liquid but redeemed before use, held for savings, interest-bearing → M2)? Step two is practical: can it be measured — is there reporting infrastructure, can double-counting be removed, can US circulation be separated from global?

Notice the asymmetry. The two bank- and fund-shaped instruments were grandfathered in by their legal wrappers. Stablecoins — the only one of the three actually being used at scale for payments — are the ones without a chair. We have written before about why tokenized deposits and stablecoins are structurally different instruments; the Fed's staff has now written down the same distinction in the language of monetary statistics.

The double-counting problem — and whose problem it is

The reason stablecoins cannot simply be added to M2 tomorrow is an accounting trap the note walks through carefully. Under the GENIUS Act, a permitted issuer must back every token 1:1 with reserve assets — and those reserves are things like bank deposits and government money-fund shares, some of which already live inside the aggregates. Add the token on top and the same dollar gets counted twice: once as the reserve, once as the claim circulating against it. The authors are direct that "the extent of the double-counting needs to be assessed to determine whether adjustments are necessary," and that the GENIUS Act's monthly reserve disclosures, while mandated, arrive in no standardized format the Fed's statistical machinery can ingest.

This is a real problem — for statisticians. For a business, the same picture contains a more practical lesson: a stablecoin is a claim on an issuer's reserve, and where the token sits determines who holds that claim. If your "crypto balance" lives inside a custodial processor, you hold a claim on the processor, which holds the token, which is a claim on the issuer — three layers deep. If the token settles to a wallet you control, you hold the instrument itself, one layer from the reserve, with no intermediary balance to freeze or fail. That difference is invisible in the aggregates and decisive in practice; it is the entire argument we laid out in our analysis of the counterparty risk merchants actually carry.

The test that decides M1 is happening at your register

Here is the sentence in the framework that deserves more attention than it got. Stablecoins would enter M1 — the money you spend, not the money you park — if they come to be "used to support everyday payment activity." The authors also concede the honest difficulty: "few payment stablecoins are currently in operation, making the assessment of their functional use difficult." Translation: the Fed's statisticians cannot yet tell whether stablecoins are payments money or parked money, because the behavioral evidence is still being generated.

Generated by whom? Not by the issuers, whose job ends at mint and redeem. Not by the exchanges, whose volumes the note would classify as trading, not payments. The evidence that would tip a stablecoin from store-of-value into medium-of-exchange is acceptance: a café taking USDC at the counter, an exporter invoicing in USDT, a SaaS charging subscriptions in stablecoins, an agency paying contractors in them. Every one of those transactions is a data point in precisely the question the Fed's staff says it cannot yet answer.

That reframes the whole news cycle around this note. The BIS argued in August that stablecoins are not credible money; the FASB proposed rules on whether they count as cash on your books; now the Fed's staff is designing the shelf they would occupy in the money supply. Three institutions, three months, one direction of travel — each treating stablecoins as something that must now be classified rather than dismissed. The classification will follow usage. It always does; the note itself records that monetary aggregates were redefined before, in 1980 and 2020, when behavior changed faster than definitions.

The two clocks: the token settles in seconds, the claim redeems in days

Buried in the note's liquidity discussion is the most operationally useful fact in it. Under the OCC's proposed rulemaking implementing the GENIUS Act, an issuer satisfies "timely redemption" if it converts tokens to dollars within two business days of a request — automatically extendable to seven calendar days whenever redemption demands exceed 10% of outstanding issuance in a single 24-hour period. The M2 club that stablecoins might join requires one-to-two-day redemption of its members; stablecoins would fit right in.

But notice what is being measured: the speed of exiting the instrument, issuer-side. The speed of using it is a different number entirely. A USDC transfer confirms in about 0.4 seconds on Solana and about 2 seconds on Base or Polygon, any hour of any day, with on-chain finality — no reversal window, no chargeback. The token runs on the fast clock; the redemption claim runs on the slow one, and the slow clock gets slower exactly when everyone is running for the exit at once.

The operational conclusion for a business writes itself. Build your payment flow on the fast clock and treat the slow one as a tail risk, not working capital. Receive the token directly to your own wallet; pay suppliers, contractors and affiliates onward in the same token; convert between stablecoins at acceptance time if you prefer one issuer's risk over another's — rather than parking value with intermediaries whose obligations to you are governed by redemption windows. A merchant who never needs the issuer's redemption desk in their daily cycle has quietly stepped outside the scenario the seven-day clause exists for. Which token to standardize on is its own question — we compare the trade-offs in USDT vs USDC for payments.

How Payzum fits the picture the Fed just drew

Payzum is a non-custodial, crypto-only payment processor, and the Fed note is close to a technical description of why that architecture exists. Every payment settles directly to a wallet the merchant controls — Payzum never holds, pools or redeems funds, so there is no processor balance layered between you and the instrument. Optional auto-convert turns whatever a customer pays into USDC or USDT at acceptance, so volatility never reaches your books. And because settlement is on-chain, every sale runs on the seconds clock: final when confirmed, with no chargebacks.

Getting there is configuration, not a project:

1. Create a merchant account and connect the wallet addresses you control — per chain, per token. This is where every payment will land.
2. Pick your acceptance surface: hosted checkout, no-code payment links and buttons, invoices with expiry and overpayment detection, recurring subscriptions — or the POS for in-person sales, where any phone becomes a terminal showing a fresh QR per sale, with PIN-scoped cashiers and per-terminal analytics.
3. Set auto-convert to USDC or USDT if you want every settlement in dollar-denominated tokens regardless of what the customer pays with.
4. Wire it into operations: signed webhooks confirm each payment to your systems, and mass payouts (CSV batches, plus EVM stablecoin payouts on Polygon, Arbitrum, Optimism, Base, BNB and Avalanche) move funds onward without touching the redemption desk.

Where the money supply question meets real businesses

Three concrete versions of "everyday payment activity" — the exact behavior the Fed's classification is waiting on:

• An importer invoicing overseas buyers: a wholesale distributor bills a foreign client in USDT via a Payzum invoice; the payment lands in the distributor's own wallet in seconds instead of a 2–5 day correspondent chain, and the payable to their own supplier goes out the same afternoon as a stablecoin payout — token in, token out, no redemption dependency.
• A counter business taking QR payments: a retail store runs Payzum POS on the phones it already owns; each sale shows a one-time QR, settles to the owner's wallet with auto-convert to USDC, and closes with zero chargeback exposure — the store's ledger is spendable dollar tokens, not a processor balance.
• An API provider selling to machines: a data company publishes an x402 URL through Payzum in front of its existing endpoint; AI agents pay per call in USDC on Base, settling straight to the company's wallet. If "everyday payment activity" ever includes software paying software, this is what it looks like.

What the Fed measures vs what your business measures

Objections worth taking seriously

"If stablecoins aren't officially money, should my business be accepting them?"

Don't confuse a statistical category with a legal or practical status. Plenty of instruments businesses accept daily sit outside M1 — and the direction of institutional travel is unambiguous: a federal reserve-backing statute in force, OCC rules in progress, FASB drafting the accounting, and now Fed staff designing the measurement. What a merchant should actually verify is narrower and checkable today: that the payment settles to a wallet they control, that the token they keep is one they have chosen deliberately, and that their invoicing, tax and AML obligations — which no rail changes — are being met.

"Should I wait until they're formally in M2 before touching this?"

The note describes reclassification as something that historically follows behavior — the aggregates were redrawn in 1980 and 2020 after usage had already shifted. Waiting for the statistical blessing means adopting at the point of maximum competition and minimum advantage. The businesses gaining something now are the ones for whom the rail solves a live problem: a corridor cards don't serve, a chargeback bleed, a settlement delay that eats working capital. If none of those describe you, you lose nothing by waiting; if one does, the M2 committee schedule is not the constraint that should decide.