For most cash programmes, the stablecoin question is how you pay your provider

By Thomas Byrnes
• • 16

What cash teams and treasurers need to ask each other before anyone runs a pilot

When cash people hear "stablecoin", we tend to picture a household with a crypto wallet. Then we argue about phones, literacy and cash-out, and the conversation stalls before it has reached the part that matters.

For most cash programmes that is the wrong place to start. The useful question is how you pay your provider.

I made that case this week to a community of practice of humanitarian cash practitioners. One participant asked how more multipurpose cash is meant to work in places where sanctions apply or the financial system is very weak. I answered with a question of my own. If you cannot pay a financial service provider to hand out cash, how are you paying the water trucking company?

In 2014 I ran a cross-border food programme into Idlib, and people used to tell me I did not do cash. I was paying bakeries, warehouses and staff every day, and all of that was money that had to arrive somewhere and be usable on the day the bill fell due. Paying a provider to distribute cash is the same kind of payment. It goes to a supplier, and it either arrives in time or it does not.

Two rooms, one gap

In July I took an early version of this work to the Treasurers' Roundtable in Washington. The message from that room was that treasurers are being asked about stablecoins before they feel ready to answer, and that what they want most is real cases from the aid sector.

Put that next to this week's session. Finance is being asked a question it cannot yet answer, and programme teams are asking about wallets when the problem sits upstream in treasury. Each side is waiting for the other to explain it.

I wrote two resources to help close that gap. Stablecoins: A Field Guide for Treasurers helps an organisation decide whether a route is worth looking at and when to stop. The Stablecoin Treasury Operational Toolkit has the working documents to test one. The Algorand Foundation commissioned both from MarketImpact and they are free to use. Neither recommends a token, a chain or a provider.

This piece is the short version for the programme side. It is what I would want a cash lead to know before they walk into finance.

Who needs usable value?

Start by sorting the request, because a stablecoin can enter a payment route in several places and they are different decisions. The question that separates them is who needs usable value at the end.

Sometimes it is your own organisation, because a donor wants to give you tokens. Sometimes it is your country office, which headquarters needs to fund. Sometimes it is a local partner or a payment provider you have to pay. And sometimes it is the person receiving assistance. Each of those has a different owner inside the organisation and a different set of approvals, and most of the confusion I hear comes from treating them as one thing. For multipurpose cash the rule I would hold on to is to separate funding the programme from paying households.

Funding the country office and paying partners or providers are where I would look first. Headquarters converts into a token, the value moves, and a provider converts it into money the response can spend. The value may sit in the token for minutes. Your country office may never hold one at all. It receives local currency and carries on under its ordinary finance controls. Holding a balance for weeks is a separate decision with its own risks and its own approval.

The stablecoin is the asset. The wallet is a delivery mechanism.

Households can be paid in stablecoin, and it has been done. In December 2022 UNHCR paid a small group of displaced people in Ukraine in USDC, straight into a wallet on their phone. UNHCR on the Ukraine pilot

So I am making a narrower claim than "never". A household wallet is a delivery mechanism, like a mobile money account or a bank card, and what sits inside it is a separate choice. Choose the modality first and then run the checks you would run on any digital delivery mechanism. What ID does it need, who controls the phone, can people cash out safely and at a fair cost, and who helps when a payment fails? If those checks fail for mobile money, expect most of them to fail for a stablecoin wallet too, unless the wallet service fixes the specific thing that failed. In Ukraine the cash-out ran through MoneyGram counters, so the pilot depended on that network being there.

Then think about what you are asking the household to hold. A dollar balance can be what people want. If the local currency is losing value quickly, or people are on the move across borders, holding dollars may protect them. Ukraine in 2022 is a fair example of the first. The central bank fixed the official rate at 29.5 hryvnia to the dollar in February, a floating cash rate ran alongside it, and in July the bank devalued by 25 per cent. I wrote about what that did to transfer values at the time. UNHCR's own launch account made the second point, that people could carry the value with them inside Ukraine or across a border. UNHCR launch account But it also hands the exchange rate to the household. Your treasury has a bank, a provider and a contract, and the household has a phone and whoever is buying dollars in the market that morning. That should be a programme decision, taken with the people receiving the money. It should not arrive as a side effect of how treasury chose to move the funds.

One more check. If people can only spend the balance at certain shops, you are running something much closer to a voucher than cash, and it should be designed and costed as one. The liquidity and redemption test I set out last year is the longer version of that check.

In most programmes you can get the access and the better rate upstream and leave the household's payment exactly as it is.

Find where the payment stops

Find out why the payment is failing before anyone books a provider demonstration. Take one real payment and trace it. Who is being paid, for what and where, and where did the last attempt get stuck?

A new route is worth testing when the payment is blocked, when it is slow, or when the exchange rate eats it. Blocked usually means de-risking. Your bank will send the money and the receiving bank will take it, but a bank in the middle has withdrawn from that corridor, so a lawful payment has nobody to carry it. I wrote about how that happens in The assumptions we built aid around. Sanctions set the legal line, and each bank then decides how far from that line it wants to stand. The Field Guide describes it as the transfer that has sat with a correspondent bank for three weeks while nobody can tell you if or when it will move. An eligible provider may be able to offer another route. Slow means the money arrives days or weeks after the invoice. And the rate problem is the one that hides, because the money arrives on time and the partner is still short. I first wrote this up in Yemen. In August 2017 a dollar cost 340 rial in Sana'a, and a fuel trader reported paying over 500 for an electronic dollar he could use abroad. It was the same country in the same month, with two rates.

Sometimes the trace finds none of these, and the grant turns out to be sitting in someone's approval queue. To be honest, that is a good result, because you have found the delay and saved yourself a pilot. The Field Guide treats stopping as a sound decision and I mean it.

Compare against the best conventional route available to you, which may not be the one you are using now. If a different bank or transfer service fixes the problem, take that and move on.

When to stop

Some routes should never reach a pilot, and the guide is firm about which.

The first is a prohibition. Sanctions and financial crime rules apply to a token as they apply to a wire. In the United States, OFAC states that the obligations are the same, and you should check every regime that applies to you. If the person you want to pay is sanctioned you cannot pay them, and if local law bans the conversion that is the answer. Changing the route cannot make a prohibited payment permissible, and no amount of risk appetite overrides a prohibition. OFAC FAQ 560 Binaya Dhakal, Global Treasurer at Mercy Corps, put the principle simply in the Field Guide. The only difference is the mode of payment, and all existing policies, procedures and control requirements continue to apply.

The second is the off-ramp, which is the point where the token becomes money your partner can spend. If nobody can lawfully do that conversion where you need it and at the size you need, the route fails. It is usually the weakest link in the chain.

The third is the donor. Read the award, because some restrict payment instruments and some do not, and for payouts get written approval or a documented non-objection before you move anything.

The reverse also holds. Permission is not the same as provision. A route can be lawful and a bank or provider can still decline to carry it, so get the provider's position in writing.

I would add one from experience. In 2025 I worked with the South Sudan Cash Working Group, through CALP, on adapting cash to exchange rate volatility. The lesson I took from it applies here. Where the authorities fix the rate by law you have a political problem, and a new payment rail does not solve a political problem.

Price what arrives

A provider quotes a percentage to move your money and your bank charges a flat wire fee, so on paper the bank looks cheaper.

Tool 3 in the Toolkit has a worked illustration with invented figures. Send USD 100,000 by bank, with USD 200 in fees and a rate of 1,000 local units per dollar, and the partner gets 99,800,000. Send it by a stablecoin route, with a fee of USD 3,000 and a rate of 1,200, and the partner gets 116,400,000. Now run both at 1,000. The stablecoin route delivers 97,000,000 and the bank wins. Operational Toolkit, Tool 3

The rate decided that comparison, and it is the number finance teams are least likely to have laid side by side. The Field Guide runs the same example and adds what the ledger shows, which is a fee fifteen times larger. The fee is in the books. The local currency you did not lose is not.

One condition sits inside that example and it matters. In the guide's version, counsel has already confirmed the route is permissible in that jurisdiction. A gap of that size usually means an official rate and a market rate are both in play. Whether you may lawfully transact at the market one is a question for counsel before it is a question for treasury.

I have seen how large that gap gets. In South Sudan the official rate moved from about 838 to about 4,464 pounds per dollar between April 2023 and March 2025, and the parallel rate from 865 to 5,750. The gap between them ran at 15 to 40 per cent. Agencies were adding a 30 per cent "inflation buffer" to transfer values and nobody could tell us how it had been calculated. When we looked, most of what it was covering was the exchange rate gap. The loss was real and it had been sitting in the programme budget under another name.

So compare local currency delivered. Get two dated quotes for the same amount, currency and destination, and ask what is firm and for how long. Ask how much the provider can pay out at that size on the day you need it, because a provider that handles a small test may not handle a month of grants. And stop the clock when the partner can pay the invoice, since a transaction can show as complete while the money waits for someone to convert it.

The partner has tokens and still cannot pay

This is the failure I would rehearse. A capped pilot goes ahead and the token reaches the partner's wallet. Then the provider says it cannot deliver the local currency before Friday, and the partner has an invoice due and a balance it cannot spend.

A good pilot has already answered what happens next. The payout was tested at pilot size before the transfer, and the amount waiting for cash-out is capped. A named person in treasury owns the incident and someone independent of the payment can halt further transfers. There is an agreed way to return the funds, and finance checks for double payment before releasing a replacement by bank. Everyone knows who bears the fees and the exchange loss, and the partner knows who to call and what it can tell the supplier.

Walk it through before any money moves, with someone playing the partner and someone playing the provider. You find out very quickly which of those answers exist.

Can you explain it?

Kathryn Powers, Head of Global Treasury at World Vision International, gave us the line I use most from the Field Guide: "If we cannot explain it, we should not use it."

For a payment route, explaining it means being able to say in plain words who holds the funds at each step, who sets the rate, who screens the parties, who proves the payout happened and who reconciles it. A provider may do several of those, or subcontract them, and that is fine as long as you can name who. Include the issuer in that. A dollar token keeps the dollar exposure, and the issuer keeps control. Circle's terms for USDC allow it to block addresses. USDC terms

The last two matter more than people expect. A transaction reference shows that tokens moved between two addresses. It does not show who approved the payment, what it was for, or that your partner received money it could use. Your auditors will want the approval, the funding record, the rate that was applied, proof of receipt and a ledger that ties them together, so agree that file before the pilot.

What a case study can and cannot tell you

Treasurers asked for real cases and there are published ones, which I documented for the guide. They need careful reading.

The public evidence is uneven. The strongest of it is published by the agencies themselves and sits in humanitarian cash and in accepting donations. Much of what is public on paying partners and providers comes from the providers who ran the pilots. A provider's account is useful, and it is also first-party.

I have quoted figures like these myself. In January, in Treasury Rails, I cited 34 per cent more purchasing power from one Sudan pilot. That was one corridor, measured against the official rate, and it does not transfer.

Here is one, read the way I would read it. Coala Pay has published a case study of a pilot with NRC in Somalia, paying pre-authorised grants of USD 6,270 to three local partners. It reports about two minutes from trigger to partner account, against an eight-day floor for the fastest existing mechanism. That is the token leg. On the local currency leg, two partners had bank deposits within 30 minutes and the third waited up to a day because it hit a Somali banking cut-off. This is the provider's own account. It is not NRC's policy and nobody independent has validated it. But look at what it shows. The clock that mattered ran to the bank deposit, and the delay came from a cut-off the partner would have known about. Coala Pay case study

A case shows that one organisation moved value along one route under its own approvals. Ask which part of the route the source describes and which part it leaves out. Another organisation's pilot is never your approval, because their donor, their bank and their destination are not yours. And where the public record has a gap, leave the gap. Do not fill it with a savings figure in a proposal.

What to take to finance

If you run cash and you think a stablecoin route might help, you do not need to become a treasury expert. You need to arrive with the right material.

Say which use case this is. In most cash programmes it is paying a partner or a provider, and the household never sees a token. Bring one real payment and show where it stopped, with the record of the last attempt, and bring the partner into the room, because they will know about a cut-off or a payout limit that headquarters finds out about too late. Then put the hard questions on the table together. Is it lawful at both ends, is there an off-ramp, does the donor allow it, and can we explain it?

Finance brings what you cannot, which is what the house bank will fund, what the delegation of authority allows and how the payment will be recorded. Between you that is the whole assessment, and the Toolkit gives you the forms for it.

Then ask the partner the question I started with. How are you paying the water trucking company?

MarketImpact advises organisations on these decisions. We do not sell stablecoin services. If I have something wrong here, tell me and I will correct it.

Tom

Thomas Byrnes is CEO of MarketImpact and works on humanitarian cash, payments and operational research.

How AI was used: AI helped draft this article from the transcript and slides of my presentations, the two handbooks, my earlier posts and the files from my 2025 South Sudan work, and checked the public references against their sources. The argument, the examples from my own work and the final edit are mine. No beneficiary data was used. More at marketimpact.org/how-we-use-ai.

Enjoyed this article?

This post is from Aid and Dev Dispatches, a LinkedIn newsletter with expert analysis on humanitarian reform, AI adoption, crisis economics, and the politics of aid. Join 9,000+ subscribers.

Subscribe on LinkedIn

About the Author

Thomas Byrnes is a Humanitarian & Digital Social Protection Expert and CEO of MarketImpact.