2023's failed banks split into two very different stories — three uninsured-deposit runs at thin-capital giants, two small-bank losses from fraud and loan concentration

Summary

Five FDIC-insured banks failed in 2023, not three — the three headline failures (Silicon Valley Bank, Signature Bank, First Republic Bank) plus two much smaller community banks (Heartland Tri-State Bank of Elkhart, KS, and Citizens Bank of Sac City, IA) that failed for unrelated reasons. The three large failures shared a real, measurable pre-failure signature versus surviving banks of similar size: roughly 68-90% of their deposits were uninsured, against a ~48% median for active peer banks with $50-300B in assets, and their equity-to-assets ratios (7.3-8.2%) ran below the ~9.2% peer median. The two small failures looked different: their uninsured-deposit shares (26-27%) were not far out of line with peers their size, but both carried noticeably thinner capital cushions (equity/assets of 3.8% and 6.0% vs. peer medians of 11.5% and 8.0%) before collapsing from CEO fraud (Heartland Tri-State) and concentrated, poorly performing commercial-trucking loans (Citizens Bank) rather than a deposit run.

Verifying the population: five failures, not three

Search results and general coverage of 2023 bank failures center almost entirely on SVB, Signature, and First Republic. Querying the FDIC's own BankFind Suite API (api.fdic.gov/banks/failures, filter FAILYR:2023) returns exactly five failures for the year:

The two small failures are three to four orders of magnitude smaller than the three headline cases and are analyzed separately below against peer banks of comparable size, since a $50-300B peer group would tell us nothing about a $60M bank.

The three large failures: an uninsured-deposit and thin-capital signature

Using each bank's Q4 2022 FDIC Call Report (the last full quarter before all three failed) and comparing against a peer group of 37 other active FDIC-insured banks with $50-300B in total assets at the same date:

All three carried uninsured-deposit shares well above the peer median — SVB and Signature were roughly double it — and equity cushions below it. SVB was the most extreme on the balance-sheet side too: its securities book (mostly held-to-maturity Treasuries and agency MBS) was $117.3B, 56% of total assets, concentrated in long-duration, low-yield instruments purchased before the Fed's 2022 rate increases; unrealized losses on that book, revealed when SVB tried to raise capital, is what triggered the run. Signature's exposure ran through crypto-industry deposits following the collapse of FTX and other crypto firms in late 2022, and First Republic's problem was concentrated in low-rate mortgages made to wealthy clients that lost value as rates rose — a similar interest-rate mismatch, expressed through loans rather than securities.

This matches published academic work: Jiang, Matvos, Piskorski and Seru (NBER Working Paper 31048, published in the Journal of Financial Economics in 2024) mark-to-market U.S. bank asset values against 2022's rate increases and find an average 10% decline (~$2 trillion) system-wide, and show formally that high uninsured-deposit leverage combined with low capital and large mark-to-market losses is what makes a bank vulnerable to a self-fulfilling run even when its assets are technically solvent at book value — exactly the combination all three failed banks show relative to peers here.

The two small failures: different mechanism entirely, not a deposit run

Heartland Tri-State (Elkhart, KS, $139.4M assets, failed 7/28/2023) and Citizens Bank (Sac City, IA, $60.4M assets, failed 11/3/2023) do not fit the uninsured-deposit story. Comparing each to peer groups of active community banks in the same size range and the same reporting quarter (last quarter before each failure — Q1 2023 for Heartland, Q2 2023 for Citizens):

Uninsured-deposit shares were modestly elevated but nowhere near the peer-doubling seen at the three large banks. The one metric that stands out for both is a thin equity cushion — Heartland Tri-State ran with roughly 75% of its peer group's typical capital ratio, and Citizens Bank with roughly a third of its peer group's. But capital thinness was not the proximate cause of either failure: reporting from Bloomberg, CNBC, and the FDIC Office of Inspector General establishes that Heartland Tri-State's CEO, Shan Hanes, embezzled $47.1 million of bank funds into a cryptocurrency "pig-butchering" scam between May and July 2023 — an act of fraud, not a market or funding event — and that Citizens Bank failed because bank examiners uncovered previously unrecognized losses in a concentrated portfolio of out-of-territory commercial trucking loans. Neither bank experienced a depositor run; both were closed by regulators after the underlying asset-quality or fraud problem was discovered.

What this means for the comparison as a whole

'How much did the banks that failed in 2023 differ beforehand' does not have one answer, because 2023's failures were not one event. The three large, headline failures (accounting for $549B of the $549B in total 2023 failed-bank assets — over 99.9% of it) shared a real, quantifiable divergence from peers: uninsured deposit funding running 40-90% above the peer median and capital ratios 10-20% below it, consistent with the literature's account of a 2022-2023 rate-driven, uninsured-deposit-run dynamic. The two small failures, while economically negligible in aggregate ($200M combined, 0.04% of the year's failed assets), diverged from peers mainly on capital thinness and were driven to failure by fraud and credit-concentration losses that had nothing to do with interest-rate risk or deposit insurance coverage. A comparison that only examined the three famous banks would have correctly described the year's dominant story but silently dropped two real failures whose causes contradict that story.