Sticky Deposits, Compressing Margins: A 117-Bank Cohort Where Funding Strength Isn't Earning Its Keep
117 of 4,108 active U.S. banks (2.85% of the universe) carry core deposits above 90% of total deposits while reporting year-over-year NIM compression at 2025Q4.** The cohort skews community — 77 of 117 sit below $500M in assets — and concentrates in the central U.S., with Texas, Oklahoma, Kansas, Missouri, and Minnesota together accounting for 45.3% of membership. The structural shape: the most stable funding posture available, paired with asset-side margin erosion that the deposit franchise is not offsetting.
Pattern
The cohort is defined by two clean predicates applied to the 2025Q4 universe: a core deposit ratio above 90%, and a negative year-over-year change in net interest margin. 117 institutions clear both screens. NIM compression across the cohort runs from a median of -17 basis points to a 25th-percentile reading of -32 basis points, with a tail extending to -113 basis points at the worst observation.
Asset-band distribution is heavily community-weighted: 46 institutions sit below $200M, 31 between $200M and $500M, 27 between $500M and $2B, 10 between $2B and $10B, and 3 between $10B and $50B. Geographic concentration is pronounced. Texas alone contributes 21 members; Oklahoma adds 10; Kansas, Missouri, and Minnesota together contribute 22. The footprint is agricultural-corridor and small-metro America.
Notable members at the steeper end of the margin-compression distribution include The First National Bank of Anson, Texas (CERT 3078), with NIM down 113 basis points year-over-year on a 91.8% core deposit base; Bessemer Trust Company, New Jersey (CERT 20015), down 104 basis points at a 92.4% core ratio; and Jonesboro State Bank, Louisiana (CERT 9325), down 87 basis points at 93.2% core. None of these institutions would surface on a conventional capital, asset quality, or concentration screen.
Precedent
Direct precedent is thin. The v0.1 failed-bank corpus contains two institutions whose pre-failure surface profile rhymes with the cohort, and both require careful framing.
The First National Bank of Lindsay, Oklahoma (CERT 4134, failed October 2024) presented before failure with an elevated core deposit share and declining year-over-year NIM — the same two structural elements that define this cohort. The actual failure mechanism, however, traced to fraud-adjacent credit issues rather than the asset-side margin compression the screen is built to surface. The analog is tight in geography and business-model character (a small Oklahoma community bank in the same regional footprint as 10 cohort members) and loose in causation. It is a partial fit, useful for posture, not for mechanism.
Heartland Tri-State Bank (CERT 25851, failed July 2023) is included as an instructive contrast. The institution presented as a deposit-rich, conservative community bank up to the discovery of CEO-orchestrated cryptocurrency fraud. The lesson runs the opposite direction from the cohort's mechanism: a clean-looking deposit franchise can mask operational weakness that no margin or funding metric will detect.
Pattern matching at this corpus depth is illustrative, not statistically rigorous. The mechanism analysis below carries more analytical weight than the precedent reference for this cohort.
Mechanism
The structural exposure works as follows. A core deposit ratio above 90% implies low reliance on brokered or wholesale funding — an unambiguous strength under standard liquidity frameworks and a positive read on the L component of CAMELS. Funding stability of this kind is precisely what supervisors prize in a rate-volatile environment, and it is what the regional banking failures of 2023 most conspicuously lacked.
The complication is that funding stability does not, on its own, generate earnings. A core deposit franchise prices through the cycle with a lag; in a sustained higher-rate environment, deposit costs rise as customers reprice into money market accounts, CDs, and competitor offerings, while the asset side — particularly fixed-rate loan books and securities portfolios originated in the 2020-2022 window — repositions only as cash flows return. When NIM compresses against a 90%-plus core base, the implication is that deposit cost creep, asset yield drag, or both are outrunning the franchise's pricing power.
This is a CAMELS earnings (E) story foremost. The cohort is not flagging on capital, asset quality, or liquidity at the individual-bank level. What the screen surfaces is a posture in which the most defensible part of the balance sheet — the funding base — is no longer compensating for asset-side conditions. Sustained over multiple quarters, that pattern compresses retained earnings, slows internal capital generation, and narrows the operating cushion available to absorb credit costs if the cycle turns.
The cohort's geographic skew toward agricultural and small-metro markets in the central U.S. compounds the diagnosis. These are markets where loan demand is slower to reprice, deposit competition has intensified as larger regional players push into rural footprints, and fee income optionality is structurally limited.
Decision
Engagement positioning. The cohort is a strong fit for Klaros earnings-resilience and balance-sheet-strategy engagements. These are not banks in supervisory distress; they are banks whose strongest attribute is no longer carrying its weight. The conversation a CFO or board at a 92% core-deposit, NIM-compressing bank wants to have is about asset repositioning, deposit pricing discipline, fee income development, and capital trajectory — not about remediation. Klaros should lead with diagnostic framing rather than risk framing.
Examination preparation. Cohort members are likely to face sharper E-component scrutiny in coming exam cycles even where overall CAMELS composites remain strong. Klaros should prepare members for examiner questions on deposit beta assumptions, asset yield trajectory under base and stressed rate paths, and the realism of NIM recovery timelines embedded in budgets. Earnings narratives that lean on funding stability without addressing asset-side mechanics will read as incomplete.
Cohort tracking. The 117-member roster should be re-run quarterly. The analytical signal sharpens if NIM compression deepens or persists into a fourth or fifth consecutive quarter, and weakens if the cohort thins as repricing works through. Klaros should also flag movement into adjacent screens — particularly any cohort member that subsequently breaches asset quality or capital thresholds, where the combination would warrant accelerated outreach.
Geographic specialization. The Texas-Oklahoma-Kansas concentration suggests Klaros's central-U.S. community banking practice is the natural delivery channel.
| # | Institution | State | Asset band | Total assets | nim change yoy |
|---|---|---|---|---|---|
| 1 | The First National Bank of AnsonCERT 3078 | Texas | Under $200M | $96M | -1.13 |
| 2 | Bessemer Trust CompanyCERT 20015 | New Jersey | $500M–$2B | $1.9B | -1.04 |
| 3 | Jonesboro State BankCERT 9325 | Louisiana | $500M–$2B | $1.2B | -0.87 |
| 4 | Idaho First BankCERT 58095 | Idaho | $500M–$2B | $1.5B | -0.81 |
| 5 | Lineage BankCERT 6100 | Tennessee | Under $200M | $164M | -0.80 |
| 6 | The First National Bank in FalfurriasCERT 14095 | Texas | Under $200M | $90M | -0.75 |
| 7 | YNBCERT 4235 | Oklahoma | $200M–$500M | $249M | -0.74 |
| 8 | Citizens State BankCERT 8183 | Michigan | $500M–$2B | $527M | -0.73 |
| 9 | Community Bank and Trust - West GeorgiaCERT 25796 | Georgia | $200M–$500M | $288M | -0.69 |
| 10 | Heartland State BankCERT 3919 | North Dakota | Under $200M | $73M | -0.64 |