The Architecture Problem Nobody Wanted to Talk About
For decades, the structural absurdity of intra-African payments was accepted as an inconvenient fact of life. A Nigerian bank settling a trade payment to a Ghanaian counterpart would route that transaction through New York. A payment from Nairobi to Accra — two cities on the same continent — would touch London before it touched either destination.
This was not a technology failure. It was an architecture failure: the correspondent banking network was built for a world where the relevant financial centres were Western, where liquidity was concentrated in USD and EUR, and where African banks were peripheral participants in a global system designed around someone else's priorities.
The cost was predictable. Correspondent banking routes added three to five days to settlement timelines. FX conversion through a third-country intermediary added five to ten percent to the cost of each transaction. And because the routing was opaque — most senders had no visibility into which correspondents handled their transaction or what the FX spread was — the market could not self-correct.
PAPSS — the Pan-African Payment and Settlement System — was developed specifically to fix this architecture. Whether it has fixed it, where it is working, and what remains broken is the honest question. I want to give you an architect's answer rather than a press release answer.
What PAPSS Gets Right
PAPSS was developed under AFREXIMBANK and launched for commercial operations in 2022. The core design is sound, and several aspects of it deserve credit.
Settlement in local currencies. The defining feature of PAPSS is that it enables bilateral settlement between member country banks without routing through a USD or EUR intermediary. A payment originating in Ghanaian cedis and settling in Nigerian naira is cleared through PAPSS in local currency terms. The USD conversion happens only at the edges — if the ultimate beneficiary needs USD, the receiving bank handles that FX event locally. The correspondent banking intermediary in New York is removed from the transaction chain.
This is architecturally correct. The correspondent banking model imposed a dollar intermediation cost on every intra-African payment regardless of whether either party actually needed dollars. Removing it reduces cost structurally, not just at the margin.
Afreximbank as settlement backstop. The multilateral liquidity mechanism — where AFREXIMBANK provides intraday settlement liquidity to smooth net positions between member central banks — addresses the liquidity problem that has historically made direct bilateral settlement unattractive for smaller currencies. Without a credible liquidity backstop, a Mozambican bank settling directly with a Sierra Leonean bank faces the problem of metical-to-leone FX depth that simply isn't there. The AFREXIMBANK facility changes the risk calculation.
Central bank participation structure. PAPSS connects at the central bank level — not at the commercial bank level directly — which is the correct architecture for a cross-border settlement system. Central bank settlement is final, irrevocable, and carries the sovereign guarantee that commercial bank settlement does not. Eight or more central banks are now operationally connected, covering a growing share of West African trade volumes.
Reduced USD dependency. The systemic benefit of PAPSS, if it scales, is reducing the continent's dependency on USD as an intermediary currency for intra-African trade. That dependency creates structural FX risk for every African institution that holds cross-border payment exposure — when the dollar moves, the cost of intra-African settlement moves with it. Local currency settlement insulates participants from that risk.
What's Still Missing
Saying PAPSS is architecturally sound is not the same as saying it is operationally complete. Several structural gaps remain.
Last-mile integration with domestic payment switches.
PAPSS settles between central banks. The question of how a commercial bank's transaction actually reaches PAPSS — and how the settled funds reach the end beneficiary — is handled by domestic payment switch integration. The quality of that integration varies dramatically across member countries.
In Nigeria, the NIBSS Instant Payment system is a mature, high-throughput rail that can connect to PAPSS cleanly. In smaller WAMZ member countries, the domestic switch infrastructure is less developed, and the technical integration between the commercial bank layer and the central bank PAPSS gateway requires custom implementation work that has not been standardised. A transaction that settles at the central bank level but cannot be credited to the end beneficiary's account in under 24 hours is not delivering on the real-time promise.
Mobile money interoperability.
The populations generating the highest volumes of intra-African payments — SMEs, diaspora remittances, trader networks — transact primarily through mobile money: M-Pesa in East Africa, MTN MoMo across West and Central Africa, Airtel Money in East and Central Africa. PAPSS in its current architecture connects bank accounts to bank accounts. Mobile money wallets sit outside the current settlement perimeter.
This is not a minor omission. In markets like Ghana, Kenya, and Tanzania, mobile money accounts outnumber bank accounts by a significant multiple. A cross-border settlement system that cannot directly address mobile wallet settlement is not solving the payment problem for the majority of intra-African commercial activity. The PAPSS roadmap acknowledges this gap. Closing it requires bilateral agreements between mobile network operators, central banks, and the PAPSS governance structure — a multilateral negotiation that is moving, but slowly.
Liquidity management for smaller currency pairs.
The AFREXIMBANK liquidity facility works at the system level. It does not solve the intraday liquidity problem at the individual commercial bank level for less-traded currency pairs.
A mid-tier bank running meaningful volumes of, say, Tanzanian shilling to Cameroonian CFA transactions needs to maintain treasury positions in both currencies, manage intraday settlement exposure, and price the FX spread accurately in real time. For major WAMZ pairs — naira to cedi, cedi to dalasi — market depth is improving. For less-traded pairs, the liquidity is thin enough that commercial banks are effectively making markets on every transaction. The treasury management burden of real-time cross-border settlement in thin currency pairs is higher than the PAPSS architecture narrative acknowledges.
Dispute resolution framework.
When a PAPSS-settled transaction goes wrong — beneficiary not credited, duplicate settlement, rejected transaction due to compliance screening — the resolution process currently lacks a standardised cross-border framework. Each member country's central bank handles disputes under its own jurisdiction and process. The absence of a multilateral dispute resolution standard creates uncertainty for commercial banks building treasury and settlement products on top of PAPSS. If you cannot specify to your corporate client how a dispute will be resolved and in what timeframe, building products on that settlement rail is a risk management problem as well as a technical one.
The Corridor Problem
Not all corridors are equal, and the gap between theoretical connectivity and actual transaction volume is where PAPSS marketing diverges most from operational reality.
Nigeria-Ghana: where PAPSS actually works. This corridor has the highest PAPSS transaction volumes in the system. Both countries are WAMZ members, the naira-cedi pair has reasonable liquidity, and both NIBSS and GhIPSS have operational PAPSS integration. For SME trade finance in this corridor, PAPSS is a genuinely usable alternative to correspondent banking. Settlement is faster, cost is lower, and the transparency of local currency settlement removes the FX opacity that characterised the old correspondent chain.
Senegal-Côte d'Ivoire: functional within WAEMU. Both countries are CFA franc users within the West African Economic and Monetary Union. The shared currency removes the FX conversion problem entirely, and both central banks have PAPSS connectivity. This corridor is effectively a domestic payment that happens to cross a border — PAPSS functions well here.
Ethiopia-DRC: theoretical. Ethiopia joined PAPSS in principle, but the operational connectivity between Ethiopian central bank infrastructure, the national payment system, and the PAPSS gateway is incomplete. DRC has even less developed domestic payment infrastructure. This corridor exists on a PAPSS member map but not in any meaningful commercial transaction sense. A bank that tells a corporate client their Ethiopia-DRC payments now route through PAPSS is making a promise the system cannot currently keep.
East Africa in general: limited. The East African Community payment integration agenda has not produced the central bank bilateral agreements PAPSS requires. Kenya, Tanzania, Uganda, and Rwanda — markets with active intra-regional trade flows and relatively sophisticated domestic payment infrastructure — are not meaningfully connected to PAPSS. This is the single largest gap between PAPSS's stated pan-African ambition and its current operational reality.
The practical architecture decision for any bank: map your actual trade corridors first, then identify which ones have real PAPSS connectivity, and treat PAPSS as an optimised rail for those specific corridors — not as a universal replacement for correspondent banking across Africa.
Comparison with Regional Settlement Alternatives
PAPSS is not the only regional settlement infrastructure banks should understand.
SADC RTGS. The Southern African Development Community Real-Time Gross Settlement system provides high-value settlement across southern African markets — South Africa, Zimbabwe, Mozambique, Zambia, and others. SADC RTGS is a proven, operational system that has been processing large-value transactions for years. Its limitation is scope: it covers high-value institutional flows, not the SME and retail corridor volumes that represent the majority of intra-African payment needs by transaction count. For a bank with significant South Africa-anchored treasury operations, SADC RTGS is the relevant infrastructure. For SME cross-border payments in the same markets, it is not.
EAC Payment Integration. The East African Community has been developing payment system integration across Kenya, Uganda, Tanzania, Rwanda, and Burundi for over a decade. Progress has been slower than the political statements suggest. PesaLink in Kenya, Uganda's NPS integration, and Tanzania's TANZIPS have all developed as strong domestic systems, but cross-border direct integration between them remains limited. The CBK-led regional payment integration discussions are ongoing. The honest assessment: EAC payment integration is a credible long-term direction with a significant implementation gap from where the infrastructure is today.
SWIFT GPI. For banks with existing SWIFT relationships and correspondent networks, SWIFT's Global Payments Innovation tracker provides real-time payment status visibility and next-day settlement in major corridors. It does not solve the cost problem — correspondent banking fees still apply — but it solves the opacity problem that historically made SWIFT unattractive for smaller-value flows. For mid-tier banks managing large-value institutional flows, SWIFT GPI is a lower-risk upgrade than building new rail connectivity from scratch.
The architecture conclusion: a well-designed African bank payments infrastructure in 2026 is multi-rail by design, not PAPSS-only. PAPSS for eligible West African corridors. SADC RTGS for southern African high-value flows. Mobile money direct integrations for SME and retail volumes. SWIFT GPI for institutional flows where correspondent relationships already exist.
Practical Implications for Mid-Tier Banks
An honest PAPSS assessment leads to a practical checklist. These are the questions a mid-tier bank should be working through now.
API readiness checklist.
| Requirement | What to check |
|---|---|
| ISO 20022 messaging | Does your core banking system produce and consume ISO 20022 pacs.008/pacs.009 messages natively, or do you need a translation layer? |
| Central bank gateway integration | Has your local central bank published a PAPSS commercial bank onboarding specification? What is the integration timeline? |
| Compliance screening | Can your AML/CFT screening system process receiving-country screening in parallel with sending-country screening, without adding sequential delay? |
| Real-time FX pricing | Can your treasury system price a local currency pair and lock the rate for a specific transaction in under 200 milliseconds? |
| Settlement reconciliation | Can your reconciliation systems handle intraday multi-currency settlement cycles rather than end-of-day batch? |
Treasury implications. Real-time settlement in local currencies changes your treasury's exposure profile. Under correspondent banking, your FX risk was concentrated in the nostro account you maintained with your US correspondent — a single position in a single currency. Under PAPSS direct settlement, you are managing positions in multiple local currencies simultaneously, with intraday settlement cycles. Treasury systems need to be capable of real-time multi-currency position monitoring. Many mid-tier African bank treasury platforms were not designed for this — they were designed for end-of-day reconciliation against a small number of correspondent accounts.
Correspondent banking relationship changes. PAPSS does not eliminate correspondent banking relationships — it reduces dependency on specific correspondent routes for specific corridors. The relationship management implication: for PAPSS-connected corridors, you may be able to reduce the volume you route through your US correspondent and renegotiate the fee structure based on lower volumes. For non-PAPSS corridors, correspondent relationships remain essential. The treasury function should map which correspondents are currently used for which corridors, and identify which relationships can be restructured as PAPSS coverage expands — before renegotiating, not after.
Compliance architecture. Dual-jurisdiction AML screening — satisfying the AML/CFT requirements of both the originating and receiving country — is a requirement for PAPSS transactions, not an optional enhancement. Banks that currently rely on correspondent banks to handle receiving-country compliance screening need to build or acquire that capability before they can settle directly through PAPSS. This is a material investment in compliance infrastructure that is consistently underestimated in PAPSS adoption planning.
The Honest Assessment
PAPSS is a genuine structural advance. Removing the New York routing requirement from intra-African payments is architecturally correct and will produce real cost reductions as adoption matures. The AFREXIMBANK backing and the central bank participation structure give it institutional credibility that payments infrastructure in this region has historically lacked.
It is not, in 2026, a complete solution. The corridor coverage gaps are real. The mobile money interoperability gap is real. The domestic switch integration quality is uneven. The dispute resolution framework is underdeveloped. A bank that builds its cross-border payment strategy around PAPSS alone — without addressing the corridors it doesn't cover, the mobile money layer it doesn't reach, and the compliance infrastructure it requires — will discover these gaps through customer complaints rather than through planning.
The path forward is a payment orchestration layer that routes intelligently across rails — PAPSS where it works, correspondent banking where it doesn't, mobile money integrations where the transaction flow demands it. PAPSS is an important component of that architecture. It is not a substitute for the architecture.
Evaluating your cross-border settlement infrastructure and need an architect's view of where PAPSS fits in your rails strategy? Talk to Raj.
Raj leads Cross-Border Payments Architecture advisory at Aicura Consulting, specialising in payment system design, ISO 20022 migration, and settlement infrastructure strategy for mid-tier banks and financial institutions.