GSEs & the FHLB system — an ~$11T quasi-sovereign pool outside the FDIC lens

Built 2026-06-14. Structured data + edges: macro-gses-fhlb.json. Companion to macro-uncovered-risk-pools, macro-mortgage-reits, and macro-bank-htm-marks.

Fannie and Freddie guarantee ~$7.7T of agency MBS on ~$179B of capital (~43x), have sat in conservatorship since 2008, and face a 2025-26 release debate. The FHLB system is the ~$1.18T "lender of next-to-last-resort" the 2023 failures ran to, with a super-lien ahead of the FDIC. (Freddie's book is ~$3.6T, not $4.1T — the round '$8.2T' combined overstates it.)

1. The GSEs — an ~$7.7T book on ~$179B of capital

• Fannie ~$4.1T guaranty book + Freddie ~$3.6T guaranteed portfolio = ~$7.7T combined (Q3-FY 2025; the round $8.2T overstates Freddie, whose book is ~$3.5–3.7T).
• Combined net worth ~$179.4B (Fannie $109.0B + Freddie $70.4B, YE2025) = ~2.3% of the book (~43x leverage). Both profitable (Fannie FY2025 net income $14.4B; Freddie $10.7B) but >$100B short of the ERCF risk-based requirement.
• In FHFA conservatorship since Sept 2008 (17+ years). Treasury's senior-preferred liquidation preference ~$341B (YE2024) and grows as retained capital adds to it — ahead of common/junior preferred.

2. The release debate — a live policy/event risk

• Trump (May 21, 2025): "very serious consideration" of taking them public. FHFA Director Bill Pulte's plan keeps them in conservatorship while selling up to ~5% (partial IPO). Officials floated a ~$1T valuation (widely doubted).
• As of June 5, 2026 the plan faces "new uncertainty"; Pulte's added role as acting DNI is expected to delay the spin-off (unlikely before the 2026 midterms).
• Rate impact: removing the implicit guarantee could add ~50bp of MBS risk premium (Stanford SIEPR); total mortgage-rate impact 20-80bp depending on whether an explicit priced backstop replaces it (Zandi: +20-40bp). A 5%-float with the backstop intact may change little; a true exit would not.

3. The FHLB system — the 2023 tell

• Consolidated obligations ~$1.18T; total assets ~$3.55T; advances ~$677B (YE2025). ~93% of US banks are members; >3/4 have borrowed.
• System advances surged ~37% in Q1 2023 to a record (~$0.8–1.0T; GAO cites a ~$1.04T system peak) — large banks 97% of the jump. SVB was FHLB-SF's largest borrower ($15B at end-2022, ~$30B by failure), Signature ~$11.3B at FHLB-NY, First Republic $28.1B — all failed within weeks; the FHLBs were repaid in full (advances exit before the FDIC).
• Super-lien (contested): advances rank ahead of the FDIC, raising resolution costs — but GAO (2026) finds its practical force eroded since 2001 UCC changes. FHFA's "FHLBank System at 100" report (Nov 2023) recommends tightening the mission.

4. The aggregate pool — wired back into banks & the Fed

• Stack: ~$7.7T GSE MBS + ~$1.18T FHLB obligations + ~$2.1T agency debt + Ginnie-wrapped MBS ≈ ~$11T of agency/GSE paper with implicit/explicit federal support, largely outside the FDIC deposit lens.
• But it is "outside" only on paper: banks held ~$2.28T of MBS (Q4 2024) and the Fed held ~$2.7T (~30% of the ~$9.2T market) at its QT-era peak (~$2.2T now) — the same collateral that marks bank books (the HTM channel) and that the Fed backstops.

Synthesis

The GSE/FHLB complex is the implicit-guarantee layer under the mortgage market: ~$7.7T guaranteed on ~$179B of capital, an FHLB system that funded the 2023 failures right to receivership, and a release debate that could reprice mortgages. A rate shock prices the GSE book, the bank securities, and the Fed's MBS at once.

What is NOT asserted

• No claim Fannie/Freddie are insolvent — both are profitable; the point is structural thinness (~43x) and the release/repricing risk.
• The ~$11T aggregate is a summation of separately-sourced components (contested); no single source publishes it.
• The super-lien is contested (GAO: eroded since 2001).
• ERCF shortfall figures vary by definition; only the direction (>$100B) is solid.
• Overlay edges are excluded from the proofs.

Sources: Fannie Q3-2025 supplement; Fannie Q3-2025 release; Freddie Q4/FY2025; Treasury SPSPA amendment; FHFA conservatorship; CNN — release uncertainty (6/5/2026); Stanford SIEPR — rate impact; FHLB Office of Finance Q3-2025; FHLBanks 2025 Impact Report; GAO-26-107373; GAO-24-106957 (2023 failures); FHFA — System at 100; SIFMA agency debt; Fed agency-MBS holdings.

← Research index · structured data: macro-gses-fhlb.json · macro-gses-fhlb.md