FDIC decrypts 2023 bank run: crypto depositors are the most likely to flee, half of deposits evaporated in three days

The U.S. Federal Deposit Insurance Corporation (FDIC) released on 5/14 its Staff Studies report titled “Dissecting Depositor Flight: An Analysis of the Spring 2023 Bank Failures,” for the first time using per-transaction data from the core systems of Silicon Valley Bank (SVB), Signature Bank, and First Republic Bank to break down the Bank Run events at these three banks in spring 2023. One of the report’s most crucial conclusions: after controlling for factors such as uninsured deposits and large-depositor status, depositors related to the digital asset industry were still significantly more likely to run; this variable’s influence on Signature Bank even exceeded the effect of whether they were large depositors themselves.

FDIC Chair Travis Hill said the study “provides a highly detailed record of deposit outflows during the fastest Bank Run events in U.S. history, deepening our understanding of contemporary banking-sector Bank Run dynamics.”

Half of deposits evaporated within three business days—an unprecedented speed

The most severe Bank Run in U.S. history included Continental Illinois in 1984 losing 30% over seven business days, and Washington Mutual in 2008 losing 10% over twelve business days. FDIC’s analysis shows that the speed of the three banks in 2023 far surpassed the above records.

Within three business days from March 9 to March 14, SVB lost 50% of its domestic deposits; from March 10 to March 14, Signature Bank also lost 50%, while First Republic Bank lost 47%. On March 9 alone, SVB had net outflows of $30.2 billion, equivalent to 20% of the March 6 deposit balance—falling by 20% in a single day is itself an unprecedented level.

The report also reveals the operational paths of the Bank Run. During the Bank Run period at the three banks, 65% to 87% of the net outflow funds left via Fedwire and SWIFT wire transfers, with little reliance on ATM cash withdrawals or branch withdrawals.

Crypto and fintech depositors ran the fastest

Signet, Signature Bank’s subsidiary, is a blockchain-based real-time payments system that attracted many crypto exchanges, stablecoin-related firms, and fintech companies offering Banking-as-a-Service (BaaS) to open accounts at the bank. The report categorizes accounts with account titles containing keywords such as “SIGNET,” “BITCOIN,” “STABLECOIN,” and “CRYPTO” as “digital asset industry depositors.”

Most deposits from this category landed under what the FDIC calls “active escrow” accounts (pooled funds held by a non-bank third party, with customers able to withdraw at any time). Signature Bank’s active escrow accounts evaporated by 83% within two business days from March 10 to March 13, and saw a cumulative loss of 88% by March 17. First Republic Bank’s similar accounts declined by 35% within three business days and had a cumulative loss of 52% by March 17.

In the FDIC’s regression analysis, “digital asset industry depositors” are among the few variables that, after controlling for factors like uninsured proportion, whether the depositor was a large depositor, and the bank relationship length, still significantly increased the likelihood of a Bank Run. The report’s original text says that at Signature Bank, the influence of this factor even exceeded “whether the depositor is a large one.” This is the first time that FDIC official documentation has explicitly defined crypto-industry depositors as one of the sources of Bank Run risk.

The top 0.5% large depositors held 39% to 62% of deposits—74% ran

The report defines depositors whose balances rank in the top 0.5% at each bank as “large depositors”: SVB has about 400 people, Signature Bank about 600, and First Republic Bank about 3,000. These three groups held 39%, 62%, and 50% of deposits at their respective banks.

From March 7 to March 17, 74% of SVB’s large depositors ran, 65% at Signature Bank, and 74% at First Republic Bank. At both SVB and Signature Bank, the Bank Run likelihood for large depositors was even higher than that of the next-tier depositors in the “top 0.5% to 1%,” and the gap cannot be fully explained by uninsured proportion or industry category alone.

Insured retail depositors flowed in instead—deposit insurance effectiveness was clear

The report also provides reverse evidence for deposit insurance. At each bank, as of March 6, depositors were fully within the $250k coverage limit, and the beneficiaries were the depositors themselves’ retail deposits (consumer accounts, small business accounts, trusts, estates). During the Bank Run period, not only did they not flee, they actually saw net inflows. SVB’s fully insured retail deposits increased by 46 percentage points between March 7 and March 17, while First Republic Bank increased by 8 percentage points.

In its regression analysis, the report points out that whether deposits are covered by deposit insurance is the most influential factor for whether people “run,” among all variables included in this study—clearly greater than relationship length with the bank or the number of accounts.

Signature Bank’s uninsured proportion: FDIC’s 90%—the study re-estimates it at 72%–76%

Another noteworthy detail in the report is that it conflicts with FDIC’s own previously published figures. The public estimate at the end of 2022 that Signature Bank had “90% of deposits uninsured” was a number adopted by both FDIC supervisory reports and the New York Department of Financial Services. But in this study, after recalculating using Signature Bank’s core system data, the uninsured proportion peaks at only 72% to 76%.

The difference mainly comes from how third-party pooled accounts are handled. FDIC’s past announcements treated the entire pooled account as a single depositor, with an upper limit of $250k; the study version instead applies a pass-through treatment for deposit insurance coverage eligibility, such as mortgage-agent custody and IOLTA lawyer trust accounts held passively. It assumes that each underlying beneficial owner would be insured separately. The research team admits this assumption may not hold for all data, but for cross-bank comparisons, it produces a relatively conservative uninsured upper bound.

For regulators, this means the “uninsured proportion” before the collapse of SVB, Signature Bank, and First Republic Bank may be lower than what the market previously believed; but the related depositors still chose to withdraw due to industry characteristics, large-depositor status, or operational funding needs. What the FDIC uses the core system data of these three banks to prove is: deposit insurance can indeed stop retail runs, but it cannot stop digital asset industry customers and large depositors from hitting the Fedwire key at speed.

This article decodes the FDIC’s 2023 Bank Runs: crypto depositors run the most, with half of deposits disappearing in three days—first appeared on 鏈新聞 ABMedia.