Reviewed quarterly analysis · $10B+ Call Report population
The pandemic liquidity buffer is gone — bank cash-and-securities coverage is back where it started
Every $10B+ asset cohort shows the same shape: liquid-asset coverage of the balance sheet surged during the 2020-2021 pandemic-era reserve buildup, peaked around Q2-Q4 2021, then declined steadily through the 2022-2023 rate-hike cycle back to roughly its pre-pandemic level, where it has stayed since. The $250B+ cohort's median rose from 33.1% to a 46.1% peak and is now at 34.2% — almost exactly where it started a decade ago. This is the real, quantified, bank-level version of the funding-risk narrative the Fed's Financial Stability Report discusses only in system-wide, qualitative terms.
Direct answer
Every $10B+ asset cohort shows the same shape: liquid-asset coverage of the balance sheet surged during the 2020-2021 pandemic-era reserve buildup, peaked around Q2-Q4 2021, then declined steadily through the 2022-2023 rate-hike cycle back to roughly its pre-pandemic level, where it has stayed since. [C04] The $250B+ cohort's median rose from 33.1% to a 46.1% peak and is now at 34.2% — almost exactly where it started a decade ago. This is the real, quantified, bank-level version of the funding-risk narrative the Fed's Financial Stability Report discusses only in system-wide, qualitative terms. [C01]
Key findings
| Question | Answer | Basis |
|---|---|---|
| How liquid is the typical $10B+ bank today? | Median 23.9% cash-and-securities coverage of assets | 159 institutions [C02] |
| Did liquidity really spike after the pandemic? | Yes — every cohort peaked in 2021, well above pre-2020 levels | Cohort history [C04] |
| Has that spike fully reversed? | Yes — all three cohorts are back near their 2016 starting levels | Cohort history [C04] |
| Who is least liquid? | Third FS&LA of Cleveland (5.8%), Bell Bank (7.4%) | Bottom ranking [C03] |
| Who is most liquid? | Custody/trust banks — Depository Trust Co (98.3%), Schwab Trust Bank (95.5%) | Top ranking [C03] |
| Did this connect to anything else in the data? | Yes — the same window saw brokered-deposit reliance spike | Cross-report finding [C10] |
The cycle, cohort by cohort
| Asset cohort | 2016 median | 2021 peak | 2026 Q2 |
|---|---|---|---|
| $10-50B | 23.0% | 32.1% (Q4 2021) | 22.4% |
| $50-250B | 22.7% | 33.9% (Q4 2021) | 25.0% |
| $250B+ | 33.1% | 46.1% (Q2 2021) | 34.2% |
The pattern is consistent across every size band: a large, temporary liquidity surge driven by the pandemic-era reserve and securities buildup, followed by a multi-year drawdown that tracks the 2022-2023 rate-hike cycle almost exactly. [C04] By 2023, every cohort had given back nearly all of its pandemic-era gain. None has since regained meaningful ground — three years later, liquidity buffers sit close to where they were a decade ago, not meaningfully strengthened by the intervening cycle.
A decade of data shows the pandemic didn't create a new, higher normal for bank liquidity — it created a temporary peak that has since fully unwound.
The same window shows up in a different balance-sheet line too
The 2022-2023 liquidity drawdown documented here happened at the same time OptimaYield's companion report found brokered-deposit reliance spiking across every cohort. [C10] That is not a coincidence of two unrelated metrics moving independently — it is consistent with a single funding-stress episode showing up in two places on the same balance sheets: banks drew down their liquid-asset cushion and reached for wholesale/brokered funding over the same stretch, exactly the kind of joint movement a funding-risk assessment should expect to see together rather than treat as separate facts.
Named-institution trajectories: three different liquidity profiles
- Bell Bank — the sharpest single move in the population: a fall from 19.5% (Q2 2021) to a low of 3.6% (Q4 2022), right at the start of the tightening cycle, with only a partial recovery to 7.4% by Q2 2026. [C05] Still well below its 2021 level five years later.
- Axos Bank — rose from 7.4% (Q1 2019) to a peak of 16.0% (Q3 2024), then declined again to 9.3% by Q2 2026. [C06] A slower, later cycle than the industry-wide 2021 peak, now also reversing.
- Third Federal Savings & Loan Association of Cleveland — persistently the least liquid institution in the entire population, ranging narrowly between 5.1% and 7.5% across the full decade with no discernible trend in either direction. [C07] This looks like a stable, structural feature of this institution's balance sheet, not a new or worsening development.
Why the highest-ratio names don't mean what they might look like
Bank of New York Mellon (72.4%) and State Street Bank & Trust (62.1%) have run the highest liquid-asset ratios among traditional banks throughout the entire 2016-2026 window, not just recently. [C08] Custody and clearing banks hold securities and central-bank reserves as a core part of their business model — safekeeping client assets and settling transactions — not as a defensive liquidity cushion built in response to stress. A high ratio here is a business-model fact, the same way a captive auto-finance bank's high brokered-deposit ratio (documented in OptimaYield's companion report) reflects its business model rather than funding pressure.
Methodology
Reuses OptimaYield's existing, already-validated liquid_assets_ratio metric contract (cash and securities divided by total assets) via the shared deriveValidatedMetrics function, unmodified — confirmed, before use, to return a normal verified value for every filer type checked, including G-SIBs and custody banks (unlike cet1_ratio, which has a real, disclosed gap for advanced-approaches banks since 2020 — see BACKLOG.md). Current-quarter population: every institution in the detailed ($10B+) Call Report layer with a derivable ratio for Q2 2026 (159 institutions). Cohort history reuses the already-computed percentile series in semantic.peer_statistics (42 quarters, 2016-03-31 through 2026-06-30). Named-institution trajectories query each bank's full available history directly. Live queries: scripts/analysis-liquid-assets-population.ts and scripts/analysis-liquid-assets-history.ts.
Limitations and counter-evidence
- This is one layer of funding resilience, not the complete picture. [C09] The Fed's own funding-risk assessment also weighs wholesale-funding reliance, interest-rate risk on the securities book, and NBFI exposure — this ratio measures reported balance-sheet composition alone.
- A low ratio is not, by itself, a risk finding. Third FS&LA of Cleveland's persistently low ratio looks structural, not deteriorating; a full assessment would need to know this institution's business model and funding mix, not just this one number.
- A high ratio is not, by itself, a strength finding either — it can reflect a business model (custody banking) rather than a deliberately built defensive cushion.
- No cause is claimed for any bank's specific move beyond the described coincidence with the brokered-deposit finding, which is offered as a consistent pattern, not a proven causal link.
- Not a frozen snapshot. Figures were pulled live from the production database rather than a versioned, hashed catalog artifact.
Glossary
- Liquid assets — cash, balances due from other institutions, and securities (held-to-maturity and available-for-sale, at carrying value), as reported on Schedule RC.
- Funding risk — the risk that a bank cannot meet its obligations as they come due without incurring unacceptable losses, a central theme of the Fed's semiannual Financial Stability Report.
Related on OptimaYield
- Q2 2026 brokered-deposit reliance — the companion piece this report's cross-reference finding (C10) draws on.
- Q2 2026 uninsured-deposit concentration — this quarter's companion piece on deposit funding structure.
- The bank-profile pages for the named institutions above (not yet linked pending publication approval).
Official sources
- Federal Reserve, Financial Stability Report.
Version history
This article was revised on 2026-09-10, but before this page began archiving prior text — the earlier version is not recoverable and is not shown here.
Evidence register
Official sources and immutable artifact fingerprints used for this publication. The visible article and structured data reference the same sources.
- Official source 1 ↗
SHA-256 9d459d61fa2ac3ab4bf34f3d20727cf053f821af0e6b1cb492b8ad142d38d975 - Official source 2 ↗
SHA-256 Unavailable
Reviewed evidence, explicit limits.
Reported facts, calculations, and editorial interpretation remain distinct. Missing values are not zero. This research is not a supervisory conclusion, failure prediction, or investment recommendation. Editorial policy →