Most CRE-concentrated banks have been that way for years — and one merger explains the sharpest outlier

Direct answer

48 of 157 banks (30.6%) in OptimaYield's $10B+ detailed population exceeded the interagency 300%-of-capital commercial real estate concentration guideline as of June 30, 2026. [C01] That is not a new development for most of them: 32 of the 48 have been above the line for all 12 of the trailing quarters available — at least three straight years — and only 4 crossed it for the first time this quarter. [C03] Concentration alone did not predict elevated delinquency this quarter, but the trend in delinquency tells a different story than the snapshot: the $10-50B cohort's own CRE delinquency rate just reached its highest point in a decade. [C02, C06]

Key findings

Why the headline number alone would mislead

"30.6% of banks exceed the CRE concentration guideline" reads as a warning on its own, and a same-quarter delinquency comparison (concentrated banks are actually less delinquent than the population median) reads like a clean rebuttal. [C02] Neither framing survives contact with the history. Most of the 48 have carried this concentration level for years without it showing up as elevated delinquency yet — but "not yet" is not "never," and one specific cohort's own delinquency trend is moving in a direction the cross-sectional snapshot alone would miss entirely.

Historical context: two different structural stories, and one boom-and-bust cycle

The $50-250B cohort's near-tripling of median CRE concentration since 2016 is a structural shift, not a cyclical swing — it climbed through the pandemic, through the 2022-2023 rate-hike cycle, and is still climbing today. [C04] The $250B+ cohort tells the opposite kind of story: a real, sharp delinquency spike that peaked in mid-2024 and has been receding since — closely matching the "banks built reserves, weathered a real CRE stress episode, and are now re-competing" narrative already circulating in trade press and the OCC's own Spring 2026 Semiannual Risk Perspective. [C05] The $10-50B cohort's delinquency trend fits neither story cleanly: it is not spiking, but it is climbing steadily to a level this cohort has not reported before in this dataset. [C06]

A cohort's concentration ratio can look unremarkable for years while its delinquency rate quietly sets a new high. The $10-50B cohort is doing exactly that right now.

Named-institution case studies: three different mechanisms

The 48 flagged institutions are not one population with one story. Three specific cases show three different mechanisms behind the same "over 300%" label:

• Flagstar Bank National Assn — a merger, not organic growth. Its concentration ratio jumped from 1.04x to 5.99x in a single quarter (Q3→Q4 2022), lining up exactly with New York Community Bancorp's completed acquisition of Flagstar Bancorp on December 1, 2022. [C07] Reading that jump as five years of organic CRE growth compressed into one quarter would be wrong — it is a balance-sheet combination, the same kind of discontinuity the existing CECL-backtest and Fragility Monitor work already treats as a reset event, not a trend point. Since the merger, Flagstar's concentration ratio has fallen steadily to 3.46x — real de-risking — while its CRE past-due/nonaccrual ratio has risen almost the whole time, from near zero to 7.58% by Q2 2026. Concentration and credit quality are moving in opposite directions on the same balance sheet.
• Banc of California — organic, recent, and unresolved. No comparable discontinuity appears in its own concentration series, which has ranged between 2.6x and 3.8x for a decade. [C08] Its delinquency, however, has risen sharply and recently — from 1.1% to 4.27% in five quarters — with no external cause identified in this pass. This is flagged as an open question, not a guessed explanation.
• Live Oak Banking Co — a slow burn. Delinquency climbed steadily from 1.82% to 3.78% over three full years while concentration barely moved (4.06x-4.47x). [C09] This is the most gradual of the three patterns — no single quarter stands out, which is itself a reason it could be easy to miss in a snapshot-only view.

Persistence: who is new, and who has been here for years

Treating all 48 as equivalent "risk" would flatten a real distinction: a bank that has run at this concentration level profitably for three years is a different story than a bank crossing the line for the first time this quarter, and this table is what actually tells the two apart.

Methodology

Reuses OptimaYield's existing, already-validated cre_loans_to_total_risk_based_capital and cre_past_due_and_nonaccrual_ratio metric contracts without modification, via the shared deriveValidatedMetrics function. Current-quarter population: every institution in the detailed ($10B+) Call Report layer with a derivable concentration ratio for June 30, 2026 (157 of 159 with schedule data present). Persistence check: for each institution over the 300% threshold as of Q2 2026, count consecutive trailing quarters (up to 12) also over the threshold, stopping at the first quarter that falls below it. Cohort history reuses the already-computed percentile series in semantic.peer_statistics (42 quarters, 2016-2026) rather than recomputing distributions by hand. Named-institution trajectories query each bank's full available history directly. Live queries: scripts/analysis-cre-concentration-population.ts and scripts/analysis-cre-concentration-history.ts.

Limitations and counter-evidence

• The $250B+ cohort's statistics are noisy by construction. [C10] With only 9-18 institutions per period, single-bank moves swing the median more than in the $10-50B cohort's 100+-institution sample — its quarter-to-quarter swings should not be read as more volatile underlying behavior without checking the underlying names.
• "Elevated" delinquency is relative, not absolute. The current-quarter cross-tab compares against this population's own median for the same quarter, not a fixed regulatory or historical benchmark.
• Banc of California's deterioration has no identified cause. Flagging it as "recent and organic" is a statement about the absence of a merger-shaped discontinuity in its own data, not a diagnosis.
• No cause is claimed, and this is not a supervisory rating. The 300% interagency threshold triggers closer supervisory attention; it is not itself a finding of unsafe or unsound practice, and naming an institution here — persistent, new, or a case study — is not a statement about its safety, soundness, or examination outcome.

Glossary

• Schedule RC-C, Part I — the Call Report schedule reporting loan composition by category, including the selected CRE loan categories used here.
• Schedule RC-R, Part I — the Call Report schedule reporting regulatory capital, including standardized total risk-based capital.
• CRE concentration guideline — the interagency (Federal Reserve, FDIC, OCC) supervisory criterion flagging banks whose CRE loans exceed 300% of total capital for closer review, from 2006 guidance on commercial real estate lending risk management.
• Persistence — how many consecutive reporting quarters an institution has remained above a given threshold, used here to separate long-standing patterns from new arrivals.
• Discontinuity — a level shift in a reported ratio too large and too sudden to be organic growth, typically caused by a merger or acquisition rather than gradual lending activity.

Related on OptimaYield

• Q2 2026 uninsured-deposit concentration — this quarter's companion distributional piece.
• The bank-profile pages for the named institutions above (not yet linked pending publication approval).

Official sources

• Office of the Comptroller of the Currency, Semiannual Risk Perspective, Spring 2026.
• Federal Reserve, FDIC, and OCC, Concentrations in Commercial Real Estate Lending, Sound Risk Management Practices (interagency guidance, December 2006) — the source of the 300%-of-capital supervisory criterion applied here.
• New York Community Bancorp, completed acquisition of Flagstar Bancorp, December 1, 2022 (public merger record).