Stablecoins hit a 9% wall on remittances

The Bank of Italy sent 200 USDC through 10 international remittance corridors and found the total cost of getting that money into someone's hands ranged from 0.3% to almost 9% of the amount sent, depending on the route.

What the mystery shopping found

Researchers tracked transfers from Italy to destinations including Argentina, Brazil, South Africa, the UAE and Japan, publishing the results as Markets, Infrastructures and Payment Systems Paper No. 86. Blockchain gas fees were a rounding error in every corridor tested. The real costs came from converting euros into USDC, exchange fees on both ends, foreign exchange spreads and local banking charges to get funds out as usable currency. Settlement times swung just as widely, from about 20 minutes where instant domestic payment rails supported withdrawal, to two business days when the recipient relied on a conventional bank transfer.

The gap between the pitch and the payout

Stablecoin remittance marketing has always centered on the cost of moving tokens on-chain, and on that narrow claim it's true: a Layer-2 transfer can cost less than a cent. But nobody sending money home is buying a blockchain transaction. They're moving value between two bank accounts in two currencies, and a recipient who needs euros or rand or reais to pay rent still has to cross that fiat boundary at some point.

Every crossing adds an intermediary. A centralized exchange to convert fiat into USDC. Another exchange or broker on the receiving end to convert back. A local bank or payment processor to push it into an account. Each one takes a cut and each one marks up the exchange rate. The Bank of Italy's finding isn't that stablecoins failed, it's that today's stablecoin remittance stack usually swaps one set of intermediaries (correspondent banks) for a different set (exchanges, brokers, off-ramps) without collapsing the total number of tolls. The corridors where stablecoins won looked like the corridors where a local instant payment system let the recipient cash out directly, cutting one toll booth rather than pricing in a new one.

That's a specific, testable claim, and it undercuts the general pitch that crypto rails are automatically cheaper than Western Union or a bank wire. In corridors with thin off-ramp competition, the FX spread and local cash-out fee can eat more than a Western Union counter charge would. The paper isn't hostile to the technology. It credits always-on settlement and programmability as real advantages over legacy rails, and it notes specific corridors where stablecoins do come out ahead. The claim under test is narrower than the industry's marketing: instant and cheap on-chain, still slow and pricey off it.

What would change the math

Watch off-ramp competition, not blockchain throughput. The paper points to Europe's MiCA framework and the spread of regulated off-ramp providers as the mechanism that could narrow conversion fees over time, since gas costs were never the bottleneck. If licensed cash-out providers multiply in corridors like the ones tested and start undercutting each other on the local-currency leg, the total cost curve compresses toward the 0.3% end. If off-ramp markets stay concentrated, foreign exchange spreads keep absorbing whatever the blockchain saves, and the remittance pitch keeps outrunning the receipt.