DNS, RTGS, and Hybrid Settlement: How Payment Systems Actually Clear and Settle

Every payment system must answer a fundamental design question: when and how does money actually move between participating banks? The answer defines the system's risk profile, liquidity requirements, speed, and cost. Three settlement models dominate global payment infrastructure: real-time gross settlement (RTGS), deferred net settlement (DNS), and hybrid approaches that combine elements of both.

Understanding these models is essential for anyone working in payments - whether you are integrating with a new market's infrastructure, evaluating settlement risk, or simply trying to understand why a payment "cleared" but hasn't "settled."

Real-Time Gross Settlement (RTGS)

In an RTGS system, each payment is settled individually, in real time, with immediate finality. When Bank A sends a payment to Bank B through an RTGS system, the central bank simultaneously debits Bank A's reserve account and credits Bank B's. The transaction is final and irrevocable from the moment of settlement.

How it works: Each participating bank maintains a settlement account (typically at the central bank). When a payment instruction arrives, the system checks whether the sending bank has sufficient funds. If yes, the payment settles immediately. If not, the payment enters a queue until funds become available.

Key characteristics: Settlement is continuous throughout the operating day Each transaction settles individually (gross) Settlement is in central bank money - the safest form of settlement Final and irrevocable upon settlement Eliminates credit risk between participants

Trade-off: RTGS systems require participants to hold significant liquidity. Because every payment settles individually, there is no netting benefit - a bank sending $1 billion and receiving $1 billion still needs $1 billion in its reserve account to fund outgoing payments before incoming ones arrive.

Examples: Fedwire (US), T2/TARGET (Eurozone), CHAPS (UK), BOJ-NET (Japan), RITS (Australia), BAHTNET (Thailand)

Deferred Net Settlement (DNS)

In a DNS system, payment instructions accumulate over a defined period - hours or an entire business day - and then settle as a single net position for each participant. Rather than settling each payment individually, the system calculates what each bank owes or is owed after offsetting all payments against each other.

How it works: During the clearing cycle, the system records all payment instructions but does not move money. At the designated settlement time, it calculates each participant's net position (total sent minus total received). Banks with net debit positions fund the difference; banks with net credit positions receive the difference. Final settlement typically occurs across accounts at the central bank.

Key characteristics: Payment instructions are batched and processed together Settlement occurs at predetermined times (end of day, multiple times per day) Netting dramatically reduces the amount of liquidity needed A system that processes $100 billion in gross payments might only need $5-10 billion in actual fund movements

Trade-off: DNS introduces settlement risk. If a participant fails between the time a payment is accepted for clearing and when net settlement occurs, the entire netting cycle may need to be unwound. This is the risk that destroyed Herstatt Bank in 1974 and has shaped payment system design ever since.

Risk mitigation: Modern DNS systems use several tools to manage this risk: Bilateral and multilateral net debit caps Collateral requirements (participants pledge securities against their net positions) Loss-sharing arrangements (if one participant defaults, others absorb the loss) Survivor-pay mechanisms (the system can complete settlement even if one participant fails)

Examples: Bacs (UK), FedACH (US), SEPA SCT via STEP2 (Europe), most ACH systems globally

Hybrid Settlement

Hybrid systems combine elements of RTGS and DNS, attempting to capture the liquidity efficiency of netting while approaching the risk reduction of real-time settlement. The specific design varies, but the common principle is: use offsetting algorithms to reduce liquidity needs, then settle the netted results continuously or at very frequent intervals.

How it works: The system continuously runs offsetting algorithms against queued payments. When bilateral or multilateral offsets are found, those payments settle together in a single, final movement. Payments that cannot be offset may settle individually (like RTGS) or wait for future offsetting opportunities.

Key characteristics: Continuous or very frequent settlement (not deferred to end of day) Offsetting algorithms reduce gross liquidity requirements Settlement finality is achieved much more quickly than DNS Lower settlement risk than DNS, lower liquidity demands than pure RTGS

Examples:

CHIPS (US): The Clearing House Interbank Payments System processes roughly $1.8 trillion daily. It uses a patented matching algorithm that continuously nets queued payments. Approximately 95% of payment value settles through bilateral and multilateral offsetting, with only the residual positions requiring funded settlement. CHIPS requires participants to pre-fund a starting position, then continuously releases liquidity as offsets are found.

EURO1 (Europe): EBA Clearing's high-value system processes payments throughout the day with continuous bilateral and multilateral netting. End-of-day net positions settle across T2/TARGET accounts at the ECB.

SPEI (Mexico): Banco de Mexico's system runs a multilateral offsetting algorithm every few seconds. Payments that can be offset settle immediately through netting; those that cannot are settled gross using central bank reserves. The algorithm runs roughly every three seconds, making settlement near-continuous.

Comparing the Three Models

Where Fast Payment Systems Fit

Fast payment systems (FPS) like FedNow, Faster Payments (UK), PIX (Brazil), and UPI (India) add another dimension. These systems typically provide near-instant payment confirmation to end users, but their underlying settlement models vary:

Pre-funded: Many FPS operate on a pre-funded model where participants place liquidity in a dedicated account. Payments draw from this pool and settle in real time or near-real time. The pool is replenished periodically. Deferred with frequent cycles: Some FPS clear instantly but settle the net positions across RTGS at regular intervals (e.g., every 15 minutes, every hour). Integrated RTGS: Systems like TIPS (Europe) settle each instant payment individually across TARGET accounts - essentially RTGS for retail payments.

The design choice has direct implications for participants' liquidity management and the system's risk profile during stressed conditions.

Why This Matters

The settlement model determines:

1. When you actually have the money: In RTGS, settlement is immediate. In DNS, the money does not actually move until the settlement cycle completes - even if the payment appears to have been received.

2. How much capital your bank ties up: RTGS demands the most liquidity. DNS the least. This directly affects the cost of participating in a payment system.

3. What happens when something goes wrong: In RTGS, a participant's failure affects only its own unsettled payments. In DNS, one failure can cascade through the entire netting cycle. Hybrid systems fall somewhere between.

4. Cross-border complexity: When payments cross borders, they typically traverse multiple settlement systems with different models and timing. Understanding these differences is essential for managing treasury positions, funding requirements, and settlement risk in international payments.

Sources: BIS - Real-Time Gross Settlement Systems (CPMI Paper No. 22); World Bank - Settlement Models in Fast Payment Systems; Bank of England - RTGS and CHAPS Annual Report 2024/25.