Credit unions, up close — the per-CU tail, CRE/MBL, the 2025 failure cluster, and the corporate-CU backbone

Built 2026-06-15. Structured data + edges: macro-credit-unions.json. The deep companion to macro-uncovered-risk-pools §1. (WebFetch unavailable; figures from search surfacing of NCUA primary pages + S&P/Ceto/OIG. The system files NCUA Form 5300, not FDIC — none of this is in the bank lens.)

Beneath the reassuring ~11.26% aggregate net-worth ratio of the ~$2.4T credit-union system: a problem-CU tail that's tiny in dollars, extreme concentration at the top, a growing CRE channel, the worst failure cluster since 2018, a single-point corporate-CU settlement tier, and an insurance fund now below its operating level.

1. Baseline & the per-CU tail

• System: 4,287 natural-person + 11 corporate CUs (12/31/2025, down from 4,411); aggregate net worth 11.26%.
• Problem tail is small: 125 CAMELS-4/5 "problem" CUs hold only ~$5.5B (~0.2% of system assets); 653 CAMELS-3. Undercapitalized (<7%) counts are period-specific/disputed (7 in one Q4-2025 analysis vs "40" in an earlier S&P framing).
• The real concentration is at the TOP: Navy Federal alone ~$181B (14M+ members); State Employees' (NC) ~$53B; the top-10 ~$456B. The systemic exposure is the giants, not the failing small CUs.
• Data gap: NCUA notes unrealized losses on long-duration securities still limit balance-sheet flexibility, but no system-wide AFS/HTM dollar figure was retrievable — the biggest open number.

2. CRE / member-business lending

• MBL capped at 12.25% of assets; total MBL $80.95B (6/30/2025); average CRE = 52.72% of net worth (low on average, localized office-refi tail). Multifamily originations +50% YoY (~$9B); other CRE +34% to $30.9B — CUs growing into CRE as banks pull back.

3. The 2025 failure cluster (7-year high)

• 6 failures Apr 30–Aug 29 2025 — most since 2018, after zero in 2024; total NCUSIF cost $18.4M, 2 from internal fraud (OIG). Named: Unilever FCU (NJ, fraud), Butler Heritage (OH), Eastern Kentucky FCU, Soul Community FCU (GA, failed ~6mo after charter), Copper & Glass (PA), Members First of Maryland (~$21.9M, Aug 29).
• Discrepancy: Ceto/S&P cite the largest 2025 failure at $58.5M but the named failures top out at ~$21.9M — unreconciled (contested). 2026 continuation: People Trust Community FCU (AR) liquidated ~Apr 2026.

4. The corporate-CU tier — the layer that actually blew up

• 2009-10: U.S. Central (~$32B) + WesCorp (~$25B) conserved; corporate losses ~$30B (~⅓ of system capital) from private-label MBS; the Temporary Corporate CU Stabilization Fund charged CUs $4.8B, closed 2017.
• Now: 11 corporate CUs (Catalyst, Alloya, Corporate One, Vizo) provide settlement, correspondent funding, emergency liquidity, and FedNow rails to ~4,287 CUs — a single point of dependence with thin retained earnings (Alloya ~$44M / Catalyst ~$43M retained). Post-2010 Part 704 reforms restricted their investments, so the residual risk is concentration/dependency, not a 2009-style MBS book.

5. Delinquency & the fund below NOL

• Delinquency 103bp (Q4 2025, +5bp); net charge-offs 78bp; loan-to-share 83.2%.
• The NCUSIF held $24.1B at a 1.30% equity ratio — BELOW the 1.33% Normal Operating Level (range 1.20-1.50%); implied insured shares ~$1.78T. The buffer absorbing the failure cluster is itself under target.

Synthesis

"Well-capitalized in aggregate" is true and misleading. The structural risks: top concentration (Navy Federal ~$181B; top-10 ~$456B); the corporate-CU settlement single-point (the layer that failed in 2009-10); an NCUSIF below its operating level; a growing CRE/MBL channel; and an un-quantified securities-unrealized-loss overhang. Same defect as the bank lens — the aggregate flatters a concentrated, partly-opaque tail outside the FDIC system.

What is NOT asserted

• No claim the CU system is undercapitalized (11.26% aggregate) — the point is concentration + the corporate single-point + the under-NOL fund + the unquantified securities overhang.
• The largest-2025-failure size ($58.5M vs $21.9M) and Copper & Glass date are contested.
• No system-wide CU AFS/HTM loss figure is asserted (the main gap); corporate-CU capital figures are single-source.
• Overlay edges are excluded from the proofs.

Sources: NCUA — Q4 2025 system performance; NCUA — NCUSIF Q4 2025 results; S&P Global — CU liquidations 7-yr high; Ceto — CU failures in 2025; NCUA OIG — failure costs $18.4M; S&P — MBL/CRE foray; Federal Register — closing the TCCUSF; Wikipedia — NCUA Corporate Stabilization Program; Bankrate — biggest credit unions.

← Research index · structured data: macro-credit-unions.json · macro-credit-unions.md