Clearing vs settlement: what actually happens to a payment

Clearing is the exchange and confirmation of payment instructions: the process by which two institutions agree exactly who owes what to whom. Settlement is the transfer of value that discharges the obligation: the moment money actually moves. A payment is cleared when both sides agree on the debt; it is settled when the debt no longer exists.

The distinction sounds like back-office pedantry. It is actually where most payment risk lives, because everything that can go wrong with a payment goes wrong in the gap between the two.

The two phases, side by side

	Clearing	Settlement
What happens	Instructions are exchanged, validated, matched, and often netted	Value is transferred between accounts
Question answered	Who owes what?	Has the money moved?
What moves	Data	Money
Risk while open	Credit and liquidity exposure between counterparties	Extinguished at settlement finality
Typical machinery	Clearing houses, messaging networks, netting engines	RTGS systems, correspondent account books, shared ledgers

A domestic payment, walked through

Take a US company paying a supplier $1 million.

Sent through ACH, the payment joins a batch. The instruction is cleared when the ACH operator validates it and calculates each bank’s net position against everyone else’s. Settlement happens later, at scheduled windows, when the net amounts move between the banks’ accounts at the Federal Reserve. This is deferred net settlement: cheap and liquidity-efficient, with an open exposure window between clearing and settlement.

Sent through Fedwire, the same payment clears and settles in one motion: the instruction is validated and the full amount moves immediately and irrevocably in central bank money. That is real-time gross settlement: no waiting and no netting, each payment settled individually, at the cost of tying up more liquidity.

CHIPS, the private-sector system that carries most large-value dollar payments, runs a hybrid: payments are netted continuously against prefunded balances and released when they can settle. The economics show why netting exists at all. CHIPS settled an average of around $2 trillion a day in 2025, and every $1 of prefunded intraday liquidity supported $26 of settled value.

The same payment, cross-border

Cross-border, the clean two-phase picture dissolves. There is no shared clearing house and no single settlement moment. The instruction travels over SWIFT from bank to bank (we walk through the full journey in how SWIFT works), and each pair of institutions in the correspondent chain clears and settles bilaterally, on its own books, in its own time.

That means a cross-border payment does not settle once. It settles several times, hop by hop, as each correspondent debits and credits the nostro and vostro accounts it maintains with the next. The sender’s bank may consider the payment gone on Monday; the beneficiary may not have final funds until Wednesday. In between, the payment exists in a state that domestic systems abolished decades ago: partly settled, with different parties carrying different exposures, and no single point at which the obligation is definitively discharged.

Why the distinction matters

Three practical consequences follow.

Finality. Until settlement is final, a payment can be unwound, and whoever has acted on it carries the loss. Knowing the exact moment an obligation is extinguished is why the concept of settlement finality is defined in law, and why it anchors every risk framework in payments.

Credit exposure. Between clearing and settlement, the payee is an unsecured creditor of the payer. The most famous failure in this gap, the 1974 Herstatt collapse, happened because one leg of an FX trade settled and the other never did, and it is the reason payment-versus-payment arrangements exist. The longer the gap and the longer the chain, the more settlement risk accumulates.

Liquidity. Deferred settlement saves liquidity through netting; immediate settlement consumes it. Every payment system design is a trade between the two, and cross-border chains sit at the expensive end: value waits in pre-funded accounts at every hop, which is capital trapped in transit.

When the two collapse into one

On programmable settlement platforms, clearing and settlement stop being phases and become properties of a single event. The validation of an instruction and the transfer of value execute as one atomic operation: both happen, or neither does. There is no window of exposure because there is no gap for exposure to live in. The BIS has described this as the settlement design goal that messaging-and-reconciliation architectures approximate from a distance, made directly enforceable by programmable settlement logic.

That is the design shift under most of the new settlement infrastructure now being built, and it changes the clearing-versus-settlement question from “how long is the gap” to “what conditions must be true for the event to execute”.

Where Frame fits

Frame is the settlement layer for global finance: one integration to orchestrate payments at scale across fiat rails, stablecoins, and tokenized deposits, with compliance enforced on every transaction and settlement in seconds instead of days.

In Frame’s model, the checks that traditionally happen in clearing run inside settlement itself. Frame’s Rules Engine evaluates every transaction against its governing policies as a condition of settling, so a transfer that cannot satisfy them does not settle, and every settled transaction produces verifiable evidence that its conditions were met. The exposure window between agreeing a payment and completing it, the gap this page is about, is engineered out rather than managed.

See how a rail-neutral settlement layer works: The Frame Blueprint.