NEW YORK — Stablecoin decks still lead with chain, yield, and integrations. Those features matter only after users and integrators trust the liability under stress. The harder product is quieter: what sits in reserve, who holds it, who can redeem at par, how often an independent accountant attests, and what happens when a custodian bank fails mid-weekend.
Supervisors have already drawn that line. On 8 June 2022, the New York State Department of Financial Services published guidance on the issuance of U.S. dollar-backed stablecoins for entities under DFS virtual-currency or limited-purpose trust oversight. The letter organizes expectations around redeemability, reserve assets, and attestations. Marketing that stops at “fully backed” without those three layers leaves the product half-described.
The token is distribution. The reserve, redemption path, and attestation regime are the product.
What an attestation actually says
An attestation is a time-stamped examination of management’s assertions under a defined scope. It is not a full audit of every risk the issuer faces. The DFS press release accompanying the guidance stressed monthly examination by an independent U.S.-licensed CPA applying American Institute of Certified Public Accountants (AICPA) attestation standards. The CPA’s work typically covers whether reserve fair value meets outstanding stablecoin units as of report dates, and whether reserve conditions set by supervisors are met. It does not promise that every banking counterpart will open on Monday.
Circle’s public transparency materials draw the same line from the issuer side. On its Transparency & Stability page, Circle states that a Big Four firm provides monthly third-party assurance that the value of USDC reserves is greater than USDC in circulation, prepared to AICPA standards, alongside weekly reserve disclosure and mint/burn flows. That cadence is a product feature. Lag, scope, random-day sampling, and whether liabilities as well as assets are reconciled all change what the PDF is worth.
When “backed” still depegs
March 2023 showed why reserve composition and access matter as much as a monthly letter. After Silicon Valley Bank failed, Circle disclosed that about $3.3 billion of USDC reserves — roughly 8 percent of then-total reserves near a $40 billion book — sat at the failed bank. Secondary-market USDC traded below $1 as redemption pressure rose. A Federal Reserve Board staff FEDS Note reconstructs the sequence: public acknowledgment of trapped uninsured deposits, solvency concerns relative to Circle’s equity, and a surge in primary-market redemptions. Circle later said, after the joint federal announcement that SVB depositors would be made whole, that the $3.3 billion would be available when banks reopened and that USDC remained redeemable one-for-one in the company’s framing (Circle press room, 13 March 2023).
The episode left monthly attestations standing as useful but incomplete. Continuous knowledge of where cash sits, which deposits are uninsured, and how fast par redemption works when a banking day is closed still matter. Integrators who only filed the latest examination PDF learned that reserve quality is a living system, not a checkbox.
Design without redeemable reserves
Algorithmic designs make the reserve product clearer by failure. In May 2022, TerraUSD (UST) lost its dollar peg and, with LUNA, collapsed as an ecosystem. The Congressional Research Service brief on algorithmic stablecoins and the TerraUSD crash described a dual-token arbitrage mechanism without the cash-and-Treasury reserve stack DFS later codified for supervised USD coins. A Richmond Fed Economic Brief walked through how the algorithmic defense broke under selling pressure. Markets discovered there was no reliable path to liquid par assets at scale. That is a product design outcome.
How firms score the reserve path
Treasury teams and payment platforms that hold large stablecoin balances usually score more than ticker liquidity.
| Feature | Useful signal | Where it misleads |
|---|---|---|
| Reserve mix | Cash, T-bills, reverse repo, fund holdings with named custodians | “Cash equivalents” that hide uninsured bank concentration |
| Segregation | Assets held for holders, apart from issuer operating cash | Legal entity maze with no clear insolvency story |
| Attestation cadence | Monthly AICPA exam + public report lag under 30 days | Annual marketing deck with no engagement letter visibility |
| Redemption rights | Who can redeem at par, cutoffs, fees, jurisdiction | Retail secondary market liquidity mistaken for primary par claim |
| Banking day risk | Multiple SIIs, known deposit insurance limits, weekend ops plan | Assuming “backed” means continuous access through any bank failure |
DFS guidance effectively scores several of those rows for New York-supervised issuers. Circle’s transparency stack scores frequency and public PDFs for USDC. Neither row set removes run dynamics. Money-like liabilities still face first-mover incentives when confidence drops. If redemption is gated, delayed, or limited to large counterparties, smaller holders discover they held a different product than the landing page implied.
Issuers compete on that disclosure stack the way money-market funds compete on holdings reports. Weekly reserve pages, named custodians, and AICPA examination PDFs are working tools when redemption desks and market makers use them. They fall short when they ignore bank concentration and weekend access. Federal frameworks for payment stablecoins may add another layer of reserve and disclosure rules over time. The product lesson from DFS guidance, Circle’s transparency program, Terra’s failure, and the SVB week remains stable either way.
The lesson for payment platforms, treasurers, and wallet brands is narrow. Chain choice and yield wrappers stay secondary until reserve composition, segregation, attestation scope, and contractual redemption are documented from primary sources. When those fields cannot be filled, the firm lacks enough of the product to treat the token as cash. When they can be filled, bank concentration and report lag still need monitoring, because March 2023 showed that markets price the reserve path in real time.
Source notes
- Guidance on the Issuance of U.S. Dollar-Backed Stablecoins, New York State Department of Financial Services, 8 Jun 2022: Supports the claim that DFS-supervised issuers must offer par redemption, hold segregated reserve assets, and pass a monthly AICPA-standard CPA examination.
- Superintendent Harris Announces Guidance on the Issuance of U.S. Dollar-Backed Stablecoins, New York State Department of Financial Services, 8 Jun 2022: Supports the claim that the guidance requires monthly examination by an independent U.S.-licensed CPA applying AICPA attestation standards.
- Transparency & Stability, Circle: Supports the claim that a Big Four firm gives monthly third-party assurance that USDC reserves exceed circulation, alongside weekly reserve disclosure and mint/burn flows.
- $3.3 Billion of USDC Reserve Risk Removed, Dollar De-Peg Closes, Circle, 13 Mar 2023: Supports the claim that $3.3 billion (about 8% of USDC's then roughly $40 billion in reserves) sat at Silicon Valley Bank and became available once federal action protected SVB depositors.
- In the Shadow of Bank Runs: Lessons from the Silicon Valley Bank Failure and Its Impact on Stablecoins, Federal Reserve Board FEDS Notes, 17 Dec 2025: Supports the claim that Circle's SVB disclosure set off solvency concerns and a surge in primary-market redemptions.
- Algorithmic Stablecoins and the TerraUSD Crash, Congressional Research Service, 16 May 2022: Supports the claim that TerraUSD's dual-token arbitrage mechanism lacked the cash-and-Treasury reserve stack DFS later required for supervised USD coins.
- Why Stablecoins Fail: An Economist's Post-Mortem on Terra, Federal Reserve Bank of Richmond, Jul 2022: Supports the claim that Terra's algorithmic peg defense broke down under selling pressure.
- Ground brief: Research log and retrieval gaps.