Chapter 028: Instant Payments and Continuous Accounting

Section 6: Payment and Settlement Accounting · Chapter 028 of 100

1. Chapter opening

Instant payments compress customer service time, but speed does not determine whether interbank settlement is gross or net. Faster Payments and UPI use arrangements distinct from central-bank RTGS services. SEPA Instant is a scheme whose participants can use different clearing and settlement infrastructures. FedNow, RTP, TIPS and Pix must each be assessed under their own operating rules.

Continuous accounting means timely event capture, position monitoring, reconciliation and controlled period snapshots. It does not abolish financial close, require event sourcing, or require each customer transaction to become a separate enterprise GL journal.

2. Learning objectives

  1. Separate customer funds availability from interbank settlement.
  2. Design balanced postings for gross and deferred settlement.
  3. Reconcile an operational subledger to aggregated GL journals.
  4. Define event, settlement, value and reporting timestamps.
  5. Maintain a reproducible period cut-off without stopping a 24/7 service.
  6. Manage prefunding, limits and out-of-hours incidents.

3. Business context

A bank may update customer balances and settlement obligations instantly, then send complete controlled totals to its enterprise GL. This works if aggregation preserves currency, entity, account, booking date and lineage, and if unmatched or delayed events are visible. A realtime GL is another valid design, not a universal mandate.

Service/infrastructureCustomer-facing serviceSettlement distinction
UK Faster PaymentsRapid 24/7 availabilityDeferred net interbank settlement with system risk controls
India UPIImmediate retail transfer serviceClearing/net settlement arrangements; not RBI RTGS
US FedNow24/7 instant paymentsGross settlement in Federal Reserve accounts
US RTP24/7 paymentsPrefunded settlement arrangement, distinct from FedNow
SEPA InstantEuro instant-transfer schemeInfrastructure-specific; TIPS is central-bank settlement
Brazil PixInstant retail serviceSPI settles in central-bank money; on-us flows differ

Customer service deadlines, limits and contingency arrangements belong in the current system specification. Do not transfer one system's parameters to another.

4. Finance and accounting view

4.1 Balanced postings

Assume a 5,000 incoming transfer is settled in central-bank money before customer credit: Dr Settlement asset 5,000 / Cr Customer deposit liability 5,000. If the bank receives value but cannot allocate it, Dr Settlement asset 5,000 / Cr Unallocated receipt liability 5,000. Allocation later: Dr Unallocated receipt liability / Cr Customer deposit liability. An alert or memo hold alone has no cash journal.

If a deferred-settlement system makes a confirmed 5,000 customer credit before interbank settlement, an illustrative sequence is Dr Scheme settlement receivable / Cr Customer deposit liability; at settlement Dr Settlement asset / Cr Scheme settlement receivable. Scheme-specific prefunding or collateral movements are separate assets or obligations, reconciled independently. Do not use a fictitious cash debit before cash actually moves.

4.2 Continuous capture and periodic close

Capture unique event ID, business date, time zone, event time, receipt time, booking time and settlement reference. A reporting snapshot uses an agreed boundary and event-sequence watermark. Processing continues into the next period; late-arriving events are assigned through controlled cut-off and adjustment rules. A midnight snapshot does not require stopping the payment engine.

Reconcile accepted instructions, operational postings, scheme acknowledgements and settlement statements. GL batch totals require counts and sums, completeness checks, duplicate controls and drill-down to every event. A delayed GL batch can be acceptable under a controlled design; a delayed spendable customer balance on a confirmed instant receipt is a different service defect.

4.3 Liquidity and prefunding

Forecast out-of-hours outflows, liquidity trapped in settlement accounts, replenishment access and scheme exposure limits. Buffer size follows stress tests and the actual settlement arrangement, not a universal 10-20% of payment volume. LCR covers a prescribed 30-day stress and does not substitute for intraday monitoring. Classification as central-bank balances, due from banks or another asset follows the legal counterparty and rights; HQLA eligibility is a separate prudential assessment.

5. Product and customer impact

Customers need correct available balances, clear pending/confirmed statuses and prompt incident handling. Real-time service can be supported by a customer ledger even where the enterprise GL receives reconciled totals periodically. Product disclosures must distinguish acceptance, beneficiary availability and settlement where relevant.

6. Regulatory and supervisory view

Use the current operator rules for settlement and service availability. EU Regulation 2024/886 introduced phased instant euro transfer obligations for covered providers, with differing euro-area and non-euro-area dates; it also changed how covered providers check users against targeted financial restrictive measures. Avoid asserting that every instant payment worldwide must undergo identical per-transaction sanctions or machine-learning checks.

Operational resilience, fraud prevention, AML monitoring and prudential reporting have separate legal and supervisory bases. A continuous service still needs period-end financial and regulatory reports, controlled adjustments and sign-off.

7. Systems and data view

One possible architecture is channel → validation and risk controls → scheme gateway → atomic customer/settlement-obligation postings → settlement and position monitor → reconciliation → controlled GL aggregation → reporting snapshot. Event sourcing can preserve immutable history, but journal tables with immutable audit logs can also support these controls. Retain replay guards, sequence checkpoints and recovery proof regardless of architecture.

8. End to end process

  1. Validate the instruction and available funds. 2. Apply applicable risk and scheme controls. 3. Exchange scheme messages within service deadlines. 4. Post the customer and settlement obligation according to confirmed state. 5. Reconcile value and any prefunding. 6. Feed complete GL totals. 7. Snapshot reporting periods while new-period processing continues. 8. Resolve breaks with event-linked corrections.

9. Controls and risks

RiskControlEvidence
Lost or replayed eventDurable event ID, replay guards, completeness totalsCount/sum and replay logs
Premature cash bookingState-driven settlement postingScheme confirmation and account statement
Period misallocationTime-zone policy, watermark and late-event workflowCut-off pack
Weekend liquidity shortagePrefunding and replenishment stress testsPosition limits and contingency drill
GL aggregation omissionSubledger-to-GL proof by currency/entity/dayReconciled batch and event population

10. Practical examples

Illustrative example A: 10,000 confirmed receipts total 25m in a customer subledger. The GL feed contains 9,999 events totalling 24.995m. The missing 5,000 must be located and posted once; customer balances need not be delayed while the controlled finance feed is repaired.

Illustrative example B: An event settles at 23:59:59 UTC and reaches reporting at 00:00:03. The cut-off policy uses the confirmed economic event and watermark to include it in the correct day. Receipt-time sorting alone would misstate two daily positions.

11. Diagrams

Figure 1. Instant customer service and settlement. Instant customer service and settlement

Figure 2. Continuous processing and period controls. Continuous processing and period controls

Figure 3. Sizing instant-payment liquidity. Sizing instant-payment liquidity

12. Tables

RecordMust prove
Customer ledgerAvailable and booked balances, event uniqueness
Settlement ledgerActual cash plus valid unsettled obligations
Enterprise GLComplete controlled totals and drill-down
Period snapshotBoundary, watermark, adjustments and reconciliation
Incident packCustomer impact, recovery and corrected postings

13. Illustrative bank case study

Fictional case: the midnight double count. A failed GL feed is replayed after midnight. Customer postings were already complete, but the finance feed lacks a batch replay key and posts the totals again. The fix links batch IDs to constituent events, reconciles both days and corrects duplicate journals. This is a finance completeness and uniqueness defect, not evidence that periodic GL aggregation is inherently unsuitable.

14. BA, developer, tester and operations guidance

  • BA: Specify state-to-journal mapping, cut-off semantics and settlement mechanism per scheme.
  • Developer: Make customer postings atomic and GL feeds complete, replay-safe and traceable.
  • Tester: Exercise midnight, time-zone, late-event, replay, partial-feed and weekend-liquidity cases.
  • Operations: Monitor scheme limits, settlement positions and aged breaks with out-of-hours escalation.

15. Common mistakes

  1. Calling every instant service RTGS.
  2. Booking cash from a message acceptance rather than settlement evidence.
  3. Confusing operational customer balances with the enterprise GL.
  4. Stopping the service to take a reporting snapshot.
  5. Sizing intraday liquidity from a generic percentage.
  6. Treating a fraud alert as automatic permission to reverse final settlement.

16. Key takeaways

  1. Customer speed and settlement mechanism are separate.
  2. Continuous capture can coexist with controlled GL aggregation.
  3. Snapshots preserve cut-off without halting new-period payments.
  4. Reconciliation proves events, obligations and actual cash.
  5. Liquidity controls follow each system's funding and settlement design.

17. References and verification notes