At a glance
The alternatives to correspondent banking are four: building local banking relationships in every market, using a local-account network that pools funds behind virtual references, settling on stablecoin rails, or clearing through a correspondent institution with direct access to local payment rails and named custody. Each replaces the chain of intermediary banks with a different structure, and each moves the risk somewhere else.
Key distinction: Correspondent banking is not slow because of its messaging. It is slow because every bank in the chain holds funds on its balance sheet and releases them on its own terms. An alternative only works if it removes the balance-sheet incentive, not just the number of hops.
This comparison covers structural differences between the four models. It is not a ranking. Which one fits depends on how many markets a platform serves, whose money it holds, and how much settlement certainty its customers need.
Why platforms look for an alternative
Correspondent banking is the system in which banks hold accounts at other banks, the nostro and vostro accounts, so that a payment from one country reaches another through a chain of bilateral relationships. It still moves most of the world's money across borders. Its capacity is shrinking while the demand on it grows.
The number of active correspondent relationships has fallen by more than 20% since 2011, according to BIS CPMI, as banks withdrew from corridors and counterparties whose compliance cost outweighed the revenue. Platforms feel this as de-banking: relationships that took a year to open close in a quarter.
At the same time, the G20 has set targets for payments across borders through the FSB's roadmap: 75% of payments reaching the recipient within one hour by the end of 2027, with the rest within one business day. The FSB's 2025 progress report shows most corridors still short of that. The gap between what platforms promise domestically and what the correspondent chain delivers internationally is the reason the incentive problem keeps coming back.
Model one: local bank build-out
The platform opens its own entities, licences, and bank accounts in each market it serves, and clears domestically in every one of them. This is how the largest global institutions operate, and it removes the chain entirely.
- What it solves. Funds move on domestic rails at domestic speed, with a direct bank relationship in every market.
- What it costs. A licence and a banking relationship per jurisdiction, local compliance teams, and capital tied up in each entity. Adding a market is a multi-year project.
- Where the risk goes. Into the platform's own balance sheet and headcount. The model only works at a scale most platforms never reach.
Model two: local-account networks
A provider that already holds accounts in many markets, typically an e-money institution or a bank partner network, gives the platform virtual account references in each currency. Collections land on the provider's local accounts and payouts leave from them.
- What it solves. One integration reaches many markets quickly, without local entities.
- What it costs. Funds sit pooled on the provider's balance sheet, and customers exist only as references in the provider's ledger. The pooled account risk is inherited wholesale.
- Where the risk goes. Into the reconciliation between three ledgers: the platform's, the provider's, and the bank's underneath. That is the structure that failed in the Synapse collapse.
Model three: stablecoin rails
Value moves on-chain as a fiat-referenced token and is converted back to local currency at each end. Settlement of the token itself is near-instant and continuous.
- What it solves. The middle of the journey bypasses banking hours and intermediary banks.
- What it costs. Both ends still need a fiat leg: a local account to collect into and a local rail to pay out on. The on-ramp and off-ramp reintroduce the same clearing and custody questions the chain had, plus the reserve quality of the issuer.
- Where the risk goes. Into the fiat legs and the issuer. Stablecoins are a settlement asset, not a substitute for local clearing, which is why they do not replace the correspondent function so much as relocate it.
Model four: a correspondent institution with direct rail access
A single institution holds direct access to local payment rails in the markets a platform serves, opens named custody accounts for the platform's customers, and clears without lending against the balances it holds. The platform integrates once.
- What it solves. Domestic clearing in each market through one counterparty, with funds held in the customer's own name rather than pooled behind a reference.
- What it costs. Dependence on one institution's rail coverage and licensing footprint, which has to be global for the model to hold.
- Where the risk goes. It is removed at the source. An institution with no lending book has no incentive to hold funds, so the queue that forms inside a lending bank does not form. The specialist clearing comparison sets out the mechanics.
A structural comparison
| Dimension | Local build-out | Local-account network | Stablecoin rails | Correspondent institution |
|---|---|---|---|---|
| Rail access | Direct, per market, owned by the platform | Indirect, through the provider's accounts | On-chain, with fiat legs at each end | Direct, through one counterparty |
| Custody structure | Platform's own accounts | Pooled, virtual references | Token balances plus fiat accounts | Named accounts per customer |
| Settlement certainty | Domestic, per market | Depends on the provider's bank chain | Certain on-chain, uncertain at the fiat legs | Domestic certainty in every market served |
| Time to add a market | Years | Weeks | Days, if a fiat leg exists | A configuration change |
| Balance-sheet incentive to hold funds | The platform's own | The provider's, and its bank's | The issuer's reserve | None: non-lending, 100% reserve |
| Regulatory footprint | A licence per market | The provider's, inherited | Issuer plus fiat-leg providers | The institution's, global |
Where each fits
A local build-out fits an institution with the capital and patience to become a bank in every market, a small set. A local-account network fits a platform that needs reach quickly and whose customers can accept pooled custody. Stablecoin rails fit flows where counterparties already hold tokens and fiat legs are thin. A correspondent institution fits platforms that hold customer money across many markets and cannot afford the reconciliation risk of pooling it or the timeline of building it.
Most platforms in practice combine models, and the combination is where fragmentation comes from. The five-layer stack of collect, hold, convert, clear, and pay out is usually sourced from three or four providers. The question is which layer's risk you are willing to own.
The infrastructure decision
The chain was never the problem. Banks in the chain behave rationally when they hold funds, because holding funds is how they earn. Any alternative that keeps a lending balance sheet in the path keeps the delay. The settlement certainty a platform can promise is bounded by the incentives of the institution operating the rails.
Lorum is the correspondent institution for banks and fintechs. It provides programmable access to global clearing, named custody, and treasury, on a non-lending, 100% reserve model. For fintech and PSP platforms and digital platforms moving money across borders, that is the fourth model: direct rails, named accounts, and an institution with no reason to hold what it is asked to move.
Frequently asked questions
What are the alternatives to correspondent banking?
Four structures replace the intermediary bank chain: building local bank relationships per market, using a local-account network pooling funds behind virtual references, settling on stablecoin rails with fiat legs at each end, or clearing through a correspondent institution with direct rail access and named custody.
How can a platform move money across borders without correspondent banks?
By collecting and paying out on domestic rails in each market instead of routing through intermediary banks. That requires either its own local accounts, a provider that holds them, or a correspondent institution with direct access to the local schemes.
Why is correspondent banking being replaced?
Active correspondent relationships have fallen by more than 20% since 2011 as banks withdrew from corridors that did not pay against their compliance cost, while G20 targets now expect most payments across borders to arrive within an hour. The chain is shrinking as the expectations placed on it rise.
Are stablecoins an alternative to correspondent banking?
Partly. They replace the middle of the journey but not the ends. Funds still have to be collected into and paid out of local accounts, so the clearing and custody questions move to the fiat legs and the issuer's reserves.
What infrastructure moves money across borders?
The combination of local payment rail access, custody accounts, currency conversion, and settlement that lets a platform receive funds in one country and deliver them in another. Correspondent banking is one way to assemble it; the four models above are the alternatives.
