Reserve composition

Why it matters

A dollar token is a claim on a portfolio. Its price says nothing about that portfolio; what it holds, at what maturity and against which counterparty, determines whether par can be paid on a bad afternoon and who absorbs the loss if it cannot. This is the one part of stablecoin regulation supervisors have written in percentages rather than principles, which makes it comparable across jurisdictions and shows how differently they define a safe asset.

It also matters outside the sector. Aggregate stablecoin supply stood at $312.1bn on 25 July 2026 on DefiLlama's measure, and the mandated lists point most of that money at one instrument: short-dated US government paper. A composition rule is therefore also a demand schedule for Treasury bills.

How it works

The US statute enacted on 18 July 2025 sets a closed list of permitted reserve assets: US currency and balances at a Federal Reserve Bank, demand deposits and insured shares at depository institutions, Treasury securities maturing in 93 days or less, overnight repurchase and reverse repurchase agreements collateralised by such Treasuries, shares in government money market funds, and tokenised forms of several of those. Cryptoassets are excluded, and reserves may not be pledged or rehypothecated except for narrow operational purposes.

The Office of the Comptroller of the Currency's proposal of 25 February 2026, on which comment closed on 1 May 2026, puts numbers on diversification. Under both options it presented, an issuer must hold at least 10% of reserve assets as demand deposits or Federal Reserve balances, keep at least 30% available within five business days, cap reserves at any one eligible financial institution at 40% and daily liquidity at a single institution at 50%, and run a weighted average maturity of no more than 20 days. An issuer with $25bn or more outstanding must also hold 0.5% of reserves, capped at $500m, in fully insured deposits. Option A makes these a safe harbour inside a principles-based standard; Option B makes them binding requirements.

Two other regimes draw the lines elsewhere. The Bank of England's policy statement and draft Code of Practice of 22 June 2026, open for comment until 22 September 2026, cap a systemic sterling issuer at 70% of backing in interest-bearing short-term UK government debt and require a minimum 30% in deposits at the Bank, alongside a temporary issuance guardrail of £40bn per stablecoin. MiCA runs the other way: a non-significant e-money token issuer must hold at least 30% of the funds received as deposits in credit institutions, rising to 60% for a significant issuer under the technical standards.

Economic mechanism

Three trade-offs sit inside any permitted-asset list, and each has an identifiable payer.

Maturity is the first. A 93-day ceiling and a 20-day weighted average maturity cap the mark-to-market loss on a forced sale, and they do it by giving up the term premium. The issuer pays in foregone carry; holders receive a portfolio that can be liquidated near par.

Counterparty is the second, and it is where the two large regimes actively disagree. A bank deposit is uninsured private credit above the insurance limit, sitting underneath an unconditional promise to pay par. MiCA compels at least 30% of the book into that exposure; the OCC's preferred safe harbour requires only 10% in demand deposits or central bank balances and caps any single institution at 40%. Allan Pedersen, in SUERF Policy Note No 411 of 16 July 2026, calls the European position narrow-bank prudence without the safety net: bank-grade prudential machinery imposed on issuers with no deposit insurance, no resolution regime and no lender of last resort, whose reserves are then routed back into the uninsured liabilities of banks that do have all three.

Remuneration is the third. Deposits at a central bank are the only reserve asset with no credit or price risk, and they earn the issuer nothing under the Bank of England's design. A 30% unremunerated floor is a levy on the issuer's income statement, paid in exchange for the ability to meet redemptions promptly. The Bank revised its earlier proposals upward, to 70% in interest-bearing assets, expressly to support the viability of issuers' business models: an acknowledgement that the price of immediacy has to leave a business behind.

The rules also feed back into the price of the asset they mandate. In BIS Working Papers No 1270, Rashad Ahmed and Iñaki Aldasoro estimate that a $3.5bn stablecoin inflow, about two standard deviations, compresses three-month Treasury bill yields by roughly 0.7 basis points on impact and around 5 basis points at a peak near day 13. The effect roughly doubles when the Treasury market is already strained, and deepens from about 3.3 basis points before 2024 to about 10 basis points afterwards. The authors add that their estimates are probably a lower bound for outflows, because they are identified over a period of growth in which issuers could choose when to buy. A forced seller has no such discretion.

Participants

Issuers set the portfolio within the permitted list. Asset managers run it: most of the USDC reserve sits in the Circle Reserve Fund, an SEC-registered 2a-7 government money market fund managed by BlackRock. Custodian banks hold the assets. Registered accounting firms examine the composition, and supervisors write the percentages. The Treasury supplies the instrument most of the money ends up in, and so has an interest in the rules directing it there.

Examples

Circle's examined reserve at 31 March 2026 held $24.91bn of US Treasury securities, $40.76bn of Treasury repurchase agreements and $10.36bn of cash at regulated financial institutions, a fair value of $77.13bn against $77.05bn of USDC in circulation. The largest single line is overnight secured lending to dealers, not the bills themselves.

Tether's consolidated report for the same date shows a different philosophy: $191.77bn of assets against $183.54bn of token liabilities, with roughly $141bn of direct and indirect US Treasury exposure, about $20bn of physical gold and about $7bn of bitcoin, and a disclosed excess of $8.23bn. Gold and bitcoin are marked at market, and what absorbs their price risk is the equity buffer, not the redemption promise.

Risks and limitations

The deposit question is unresolved and the evidence cuts both ways. The clearest argument for the American position is the March 2023 episode, when USDC traded at a discount because $3.3bn of reserve cash sat at a bank that failed over a weekend. The clearest argument for the European position is that a reserve held entirely in government paper and dealer repo makes the issuer a shadow money market fund with none of a bank's obligations, and concentrates the sector's stress in the repo market. The ECB has pushed for narrower, central-bank-anchored backing; the Commission is considering allowing European government money market instruments in the MiCA review; the EBA has defended the existing text. Nobody has run the experiment at scale.

Repo concentration is a related exposure. More than half of the USDC reserve is an overnight secured claim on dealer counterparties: a strength in ordinary conditions, and the exact market in which a funding squeeze would first appear.

Finally, the rules are written on the assumption that the permitted market is deep enough to absorb the sector. The BIS estimates say that assumption is already weakening at the margin. A composition mandate that concentrates a growing sector into a single instrument makes the sector a price setter in that instrument, which is a systemic consequence of a prudential rule.

Key metrics

Watch the split between government paper, repo, bank deposits and central bank balances, and the weighted average maturity against the proposed 20-day cap. Track concentration at any one institution against the proposed 40% limit, the disclosed excess over liabilities, and the unremunerated share of backing, which measures what the par promise costs the issuer. Aggregate supply ($312.1bn on DefiLlama's count on 25 July 2026 against $297.88bn on RWA.xyz's narrower register the same day) bounds sector demand for the permitted assets, and the gap between the two counts shows that even the denominator is a definitional choice.

Related research

BIS Working Papers No 1270 is the best available estimate of how reserve flows move the price of the reserve asset, and SUERF Policy Note No 411 the sharpest critique of the European deposit requirement. The Sullivan & Cromwell summary of the OCC's February 2026 proposal is the most usable account of the American quantitative standards; the Bank of England's June 2026 statement shows a third supervisor choosing a third answer.

The data on this site

The series below are the ones that bear directly on this concept. Each carries its own source, cadence and coverage.