Federal Reserve warns GENIUS Act may not stop stablecoin runs caused by network congestion

How network fees can trigger stablecoin redemptions

A Federal Reserve staff paper warns that congestion on public blockchains can cause holders to redeem digital dollars or move them to other networks, even if the token's reserves are completely safe.

The study, updated on August 31, 2026, models how rising transaction fees reduce the usefulness of a payment token. When fees climb too high relative to the value of small payments, holders may leave. In times of low usage or weak "network effects"—meaning fewer people accept or use that specific token—these individual exits can turn into coordinated redemptions.

This finding comes as the US implements the GENIUS Act, the new federal framework for payment stablecoins. While the law tightens rules for issuers and their reserves, it does not regulate the public blockchains where those tokens live. Those chains have their own fee markets and capacity limits.

Key numbers from the research

• The threshold effect: Under high congestion and weak network effects, a $10.83 increase in Ethereum gas fees was linked to a roughly 0.9 percentage-point rise in weekly redemptions. Gas fees alone were not statistically significant in all cases; the effect appeared mainly when fees interacted with weaker network usage.
• Small payments hit hardest: Between 2021 and 2025, for small USDC transfers (below the median), the fee-to-value ratio exceeded 100% frequently at the 75th percentile. For larger transfers, fees rarely exceeded 5% of the value.
• Cross-chain moves: Analysis of matched USDT transfers between Ethereum and Tron showed that $1 more in Ethereum gas fees was associated with 3% to 4% more value moving to the alternative chain.

What the evidence confirms

The paper combines theoretical modeling with empirical data from November 2017 through December 2025. It distinguishes between different types of evidence:

• Capacity shocks: The study uses Ethereum's "empty slot" design as a proxy for congestion. Empty blocks reduce capacity and push up fees. This link is supported by strong causal design, though it does not prove every fee spike causes redemptions.
• Redemption measurements: In the data, "redemption" is measured as a drop in a stablecoin's circulation on Ethereum. This includes both cashing out to fiat and migrating to another blockchain. The data track pressure on Ethereum-based holdings rather than direct issuer cash-outs.
• Conditional warning: The authors describe the result as a "conditional warning," not a forecast. The pattern is consistent with the model but remains a historical association.

The GENIUS Act gap

The GENIUS Act gives regulators broad tools to police stablecoin issuers, their reserves, and redemption promises. However, its explicit reserve provisions and Treasury's current section 3 proposal set no price or capacity standards for public blockchains.

This means that while issuers must hold safe assets, the infrastructure moving those assets remains subject to unpredictable fee spikes and congestion. The Fed paper sharpens an unresolved question for Treasury implementation: how to protect digital dollars when the underlying network becomes expensive or slow.

Why this matters for users

For everyday users, the research highlights a risk beyond issuer solvency. Even if a stablecoin is fully backed, high network fees can make small transactions uneconomical. During congested periods, users may avoid completing transfers, wait for fees to drop, batch activities, or move to cheaper blockchains like Tron.

This creates a fragmentation risk where the same dollar-backed token behaves differently depending on which chain it sits on, potentially eroding trust in the payment utility rather than the reserve quality.

What is still unclear

Several limitations remain. The redemption data captures negative changes in Ethereum circulation but cannot distinguish every owner or motive behind each transfer. The association between fees and redemptions was strongest during periods when network effects were weak, which accounted for only about 7% to 7.5% of observations. It is unclear how often this threshold will be triggered under current or future GENIUS regulations.

Sources

• CryptoSlate