Two African businesses trading with each other have long paid a third party for the privilege. A Ghanaian importer buying from a Nigerian supplier has typically converted cedi into US dollars, routed the money through correspondent banks in Europe or North America, then converted a second time into naira. Two currency spreads, several days of float, and a settlement chain that answered to no African institution. On 13 January 2022 in Accra, Afreximbank and the African Union officially launched the Pan-African Payment and Settlement System as a direct response to that detour.
The Model: Settlement stripped of the foreign intermediary
Stripped to its logic, PAPSS is an attempt to redraw the plumbing of intra-African trade. The proposition is real-time settlement between African markets, with each party paying and being paid in its own currency while the system clears the cross-border leg centrally. The strategic claim is not merely convenience. It is the removal of a structural dependence: the reliance on third-country currencies and offshore correspondent relationships to complete an African-to-African transaction.
For an operator, the interesting question is where the value actually sits. It sits in three places at once — lower foreign-exchange cost, compressed settlement time, and reduced counterparty risk in the correspondent chain. Each of those is a line item a treasurer can model. A payment rail is only as valuable as the friction it removes.
The Logic: A public rail for a continental market
PAPSS is best read as the financial counterpart to the African Continental Free Trade Area. AfCFTA lowers the tariff and rules-of-origin barriers to moving goods; PAPSS attacks the barrier to moving the money that pays for them. One without the other is incomplete — a duty-free consignment still stalls if the payment takes a week and loses value to conversion.
The design choice worth noting is that PAPSS is infrastructure, not a product. It sits beneath banks and fintechs rather than competing with them, becoming useful only as national institutions integrate. That is the same architecture as a domestic instant-payment switch, extended across borders. The policy logic is deliberate: build the shared road, let private firms run the vehicles on it.
The Test: Which assumptions could fail
The harder analytical work is identifying where the model could disappoint. Three assumptions carry the weight. The first is liquidity — the system presumes participating central banks and settlement agents can provide enough of each currency to clear net positions without strain. The second is participation density; a payment network delivers value in proportion to who is actually on it, and coverage on launch day is narrower than the eventual ambition. The third is trust in local currencies themselves — a rail that settles in African currencies inherits their volatility, and an exporter paid in a soft currency may still prefer dollars.
The intellectual-property and governance questions are quieter but consequential. Who owns the standards, who arbitrates disputes, and how is the switch governed across sovereign monetary authorities. These are not reasons to dismiss the model. They are the variables an operator should watch as evidence accumulates. A framework is credible only where its failure points are named.
The Decision: Enter, integrate or observe
For a West African bank or fintech, the immediate choice is whether to integrate early and shape the corridors, or wait for volumes to prove out. For an exporter or importer, the decision is narrower and more testable: run a live transaction, measure the true landed cost against the dollar route, and keep the one that wins on price and speed.
The second-order effects are where the real story will be written. If settlement friction falls durably, marginal trades that were never worth the FX cost become viable, and the composition of intra-regional commerce shifts toward smaller, more frequent flows. Whether that materialises depends on integration, not intention. As of this week, PAPSS is a well-argued proposition with its proof still ahead of it — which is precisely why serious operators should be modelling it now rather than reading about it later.
