A commercial bank with several concentrations
Signature Bank’s failure is often compressed into a cryptocurrency story. That misses the structure of the bank. Its core activities were commercial real estate and commercial-and-industrial lending funded by business deposits. Digital-asset customers were one important funding constituency, alongside established commercial relationships. The relevant questions are who could withdraw, how quickly, and what assets could fund their exit.
The FDIC’s April 2023 review reports year-end 2022 assets of $110.4 billion, loans of $74.9 billion and deposits of $88.6 billion. Uninsured deposits were $79.5 billion, approximately 90% of total deposits. Digital-asset deposits were $17.8 billion, down from $28.7 billion a year earlier. These figures describe a bank with broad commercial lending and highly uninsured funding, rather than a balance sheet consisting mainly of cryptocurrency investments. [1]
Original calculations using those rounded figures put digital-asset deposits at 20.1% of year-end deposits and 16.1% of assets. The two denominators answer different questions. Calling the 16% asset ratio a deposit share understates the funding concentration; calling the 20% deposit share the share of crypto assets misidentifies the exposure entirely.
Scroll horizontally to see all columns.
| Year-end measure, $bn | 2021 | 2022 | Interpretation |
|---|---|---|---|
| Total deposits | 106.2 | 88.6 | Bank liability stock [1] |
| Digital-asset deposits | 28.7 | 17.8 | Deposits from that customer segment, not coins owned [1] |
| Digital-asset share of deposits | 27.0% | 20.1% | Original calculation from rounded inputs |
| Uninsured deposits | 97.6 | 79.5 | Dollars above applicable insurance coverage [1] |
| Uninsured share of deposits | 91.9% | 89.7% | Original calculation; reported ratios round to 92% and 90% |
Signet and the meaning of a sticky relationship
The FDIC inspector general describes Signet, launched in 2019, as a blockchain-based platform enabling commercial customers to make real-time payments to one another. Management regarded the platform as increasing the durability of deposits. The same review found that the bank relied too heavily on assumed customer loyalty and lacked sufficient risk-management practices as it grew. [2]
A useful payments platform can make an account operationally valuable without making the cash in it immobile. A business may keep an account open while moving most excess funds elsewhere. A relationship can therefore survive in a commercial sense while the funding balance supporting the bank’s assets disappears. Retention of customers and retention of dollar balances are different tests.
Nor does use of blockchain settlement itself explain the failure. The immediate obligation was to honor dollar deposit withdrawals. Digital infrastructure mattered to customer activity and reputation, but an assertion that token technology caused the run would require evidence beyond the existence of Signet.
The lending book was not the deposit customer list
The FDIC’s March 29 testimony identifies a roughly $34 billion commercial-and-industrial portfolio, including approximately $28 billion of fund-banking loans to private-equity firms and their general partners. This helps distinguish the bank’s commercial lending exposures from the digital-asset deposit segment. Its real-estate book added a different source of long-lived and illiquid exposure. [3]
A fund-banking facility can have a different credit structure and maturity from a multifamily mortgage, but both still require funding. If a customer segment withdraws deposits, the bank cannot necessarily call unrelated performing loans immediately. The label of the deposit business and the identity of the borrower need not match for the shock to transmit across the balance sheet.
This is why a run can occur without a sudden wave of borrower defaults. The receiver may later realize gains or losses on loans, but the inability to fund same-day withdrawals is a different event from proving the ultimate credit loss on those loans. Loan-sale discounts may reflect rates, liquidity, structure and servicing costs as well as credit expectations.
March 10: a flow measure that needs its own denominator
The inspector general reports $18.6 billion of withdrawals on Friday, March 10, equivalent to about 20% of the bank’s deposits at that point. This was a contemporaneous run measurement. Dividing $18.6 billion by the older December 31 balance of $88.6 billion gives 21.0%, but that is an explicitly mixed-date comparison, not a correction to the official 20% description. [2]
New York’s Department of Financial Services describes the run after Silvergate’s wind-down announcement and SVB’s closure, followed by a nearly $4 billion cash deficit requiring late emergency assistance. Over the weekend, Signature’s estimates of known withdrawals pending for Monday climbed, and supervisors lacked confidence in the bank’s ability to produce a credible funding plan. The state took possession on March 12. [4]
The distinction between completed Friday withdrawals and queued Monday instructions is essential. The latter measures prospective pressure. Adding them without qualification treats uncertain future demand as settled cash flow and can conceal how rapidly management’s estimates were changing.
Collateral readiness and the quality of management information
The state review says Signature could not provide reliable and consistent information on available and pending withdrawals during the crisis. It described management counting on collateral that would not be immediately acceptable or would need weeks of review. A liquidity calculation built from such assets did not answer the practical question of whether the bank could open safely on Monday. [4]
The institutional failure here was partly operational. Documentation, collateral eligibility and transfer procedures determine whether an asset can support a draw in the available hours. Management information also determines how much cash is required. An apparently adequate pool of assets can coexist with a deficit when liabilities are measured too slowly and funding sources too optimistically.
This does not imply that flawless preparation would certainly have saved Signature. A sufficiently large run can overwhelm a prepared institution too. It means that incomplete preparation made the actual response weaker, and that claims of inevitable survival or inevitable failure go beyond the observed record.
What supervisors knew before the final weekend
The FDIC’s April review attributed the failure to poor management and recognized weaknesses in its own supervisory response, including the potential for earlier escalation. It described a bank whose rapid expansion outpaced its controls. Those are supervisory findings, not a criminal verdict on individual conduct. [5]
The inspector general found missed opportunities to downgrade the management component and escalate concerns. Its review also identifies delayed and canceled supervisory work. Together with the state record, this supports a narrower and more useful conclusion than “regulators knew nothing”: risk signals existed, but neither remediation nor intervention kept pace with the exposure. [2]
A sustained satisfactory overall impression can coexist with a serious weakness in one component. Earnings and low realized loan losses are backward-looking evidence; contingency funding is a forward-looking capability. The case illustrates why the distinction matters without proving that a particular earlier supervisory action would have changed every later depositor decision.
What the record supports about cryptocurrency
Digital-asset deposit contraction during 2022 reduced one concentration but did not remove the dependence on uninsured commercial funds. The bank remained exposed to customers who could move large balances quickly, and public association with other stressed institutions could accelerate that behavior. A smaller crypto-related deposit share was therefore not evidence of a fully repaired funding structure.
The state’s review found that the digital-asset share of March 10 withdrawals was broadly proportional to its share of deposits. That finding does not support attributing the entire run to that segment. The stronger documented mechanism is a broad loss of confidence interacting with an uninsured funding base and weak execution. [4]
The distinction also prevents a category error after resolution. Dollar deposits belonging to a digital-asset company are bank liabilities. Tokens held by an exchange’s customers, securities owned by that company and its own corporate obligations are separate claims. Protection of the bank deposit does not imply federal protection of all those other assets.
Bridge-bank protection and the Flagstar transaction
The FDIC established Signature Bridge Bank so banking activities could continue on March 13. Deposits were protected under the systemic-risk response. Protection of customers’ deposits did not preserve the old shareholders’ investment or guarantee all unsecured obligations. [6, 3]
On March 19 the FDIC announced Flagstar’s agreement to take substantially all deposits and selected loan portfolios. The transaction involved $38.4 billion of assets, including $12.9 billion of loans purchased at a $2.7 billion discount. Approximately $60 billion of loans remained in receivership. The $2.7 billion was the loan discount, not the price paid for the entire bank or for all $38.4 billion of assets. [7]
About $4 billion of digital-asset-related deposits were excluded from Flagstar’s bid. The FDIC said it would return those deposits directly to the customers. Exclusion from the buyer’s business perimeter therefore did not mean the deposits had been denied the emergency protection. Likewise, the buyer did not acquire the entire old loan book merely because branches reopened under its name. [7]
The retained real-estate book and a second resolution stage
One later transaction illustrates the long tail. In December 2023, the FDIC receiver contributed about $9 billion of rent-stabilized or rent-controlled multifamily loans to a venture. A Santander-controlled entity paid $1.1 billion for a 20% equity interest; the receiver retained 80%. The FDIC cited its statutory obligation regarding availability and affordability of housing for low- and moderate-income individuals. [8]
The $1.1 billion cannot simply be divided by the $9 billion loan balance and called the sale price for the entire portfolio. It purchased a minority equity interest in a structured entity. Financing, retained ownership and the contractual distribution of cash flows determine the economics. Loan balance, venture equity and cash consideration are different layers.
The FDIC’s 2023 annual report separately records $392 million received after exercising and selling the equity-appreciation rights associated with the earlier bank transaction. It also reports a year-end estimated Signature cost of approximately $1.8 billion, compared with the initial $2.5 billion announcement estimate. Those are dated estimates and receipts, not evidence that subsequent recoveries or expenses were finished. [9]
Outcomes and remaining uncertainty
Depositors received continuity or payment; the former bank’s investors and nondeposit claimants had different legal positions. The FDIC’s receivership page provides the relevant claim process, while loan borrowers remain obligated under their contracts despite a change of owner or servicer. A brand transition is neither debt forgiveness for borrowers nor a universal recovery statement for creditors. [10]
The special assessment concerns the portion of SVB and Signature losses arising from uninsured-depositor protection. The FDIC states that estimates change with asset dispositions, claims and expenses. A combined assessment amount cannot be allocated to Signature alone without institution-specific supporting data. [11]
The May 2026 FDIC depositor study adds transaction-level evidence that some very large depositors withdrew almost everything even from accounts used in business operations. Signature’s distinctive lesson is that relationship depth, digital payment utility and nominal collateral value each failed to guarantee stable funding. Their value depended on customers’ confidence and the bank’s capacity to convert assets into cash quickly enough. [12]
Sources
- FDIC, Supervision of Signature Bank, April 28, 2023, financial-information tableOfficial release · PDFBack to text: ↑1↑2↑3↑4
- FDIC OIG, Material Loss Review of Signature Bank, October 2023Official source · PDFBack to text: ↑1↑2↑3
- FDIC, Recent Bank Failures and the Federal Regulatory Response, March 29, 2023Official sourceBack to text: ↑1↑2
- New York DFS, internal review of Signature supervision and closure, April 28, 2023Official source · PDFBack to text: ↑1↑2↑3
- FDIC, supervisory-review findings release, April 28, 2023Official releaseBack to text: ↑
- FDIC, Signature Bridge Bank establishment, March 12, 2023Official releaseBack to text: ↑
- FDIC, Flagstar purchase-and-assumption announcement, March 19, 2023Official releaseBack to text: ↑1↑2
- FDIC, rent-regulated multifamily venture transaction, December 20, 2023Official releaseBack to text: ↑
- FDIC, 2023 Annual Report, receivership discussion p. 81; Federal Reserve Bank of St. Louis archiveFiling / report · PDFBack to text: ↑
- FDIC, Signature Bank receivership informationOfficial sourceBack to text: ↑
- FDIC, Special Assessment Pursuant to Systemic Risk Determination, checked October 4, 2026Official sourceBack to text: ↑1↑2
- FDIC, transaction-level depositor-flight research release, May 14, 2026Official releaseBack to text: ↑