Mortgage REITs & REITs — the levered, repo-funded MBS holder that already fire-sold in 2020

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

Mortgage REITs borrow short in repo to hold long-duration agency MBS at ~6-9x leverage — the same MBS banks (~$3.0T) and the Fed (~$2.3T) hold. A rate/spread shock forces margin-driven selling into that shared market; in March 2020 Invesco Mortgage went from ~$22B to ~$1.6B in ~10 weeks, and only unlimited Fed MBS buying stopped the cascade.

1. The model & the rate-shock pipe

• Borrow via repo (rolled <90 days) to hold long-duration agency MBS, hedged with swaps. Annaly Q1 2026 economic leverage 5.7x (GAAP 7.3x); AGNC Q3 2025 "at risk" leverage 7.6x.
• Rising long rates + MBS spread widening crush book value; repo rollover + margin calls force MBS sales into the same market held by banks (~$3.0T) and the Fed (~$2.3T, ~30% of the ~$9.2T agency-MBS market) — marking down bank AFS/HTM simultaneously.

2. The two giants & their 2022 drawdowns

• Annaly (NLY): ~$107B portfolio (~$92B liquid agency); BV/sh $19.82 (Q1 2026). In 2022, BV/sh fell ~15% in Q1 and ~18% in Q3.
• AGNC: ~$91B portfolio (95% 30yr fixed); leverage 7.6x→7.2x. 2022 economic return −28.4%; tangible BV/sh fell $15.75 (YE2021) → $8.70 (YE2023), −45% — 2022 was the worst year for the US MBS index since 1976.

3. March 2020 — the proof-of-concept

• Invesco Mortgage (IVR): after meeting calls through Fri Mar 20, 2020, on Mon Mar 23 it could not fund that day's margin calls. It sold $16.2B of MBS/CRT and repaid $11.2B of repo in Q1; the portfolio collapsed $21.9B (YE2019) → ~$1.6B (5/31/2020).
• The cascade blew out spreads until the Fed announced unlimited MBS QE on Mar 23, 2020. The explicit precedent: the Fed is the buyer of last resort in the same MBS market that marks the banks.

4. Commercial mREITs — the CRE-credit channel

• BXMT cut its dividend 24% (to $0.47, Q3 2024), ~36% office (2024); STWD holds ~$10B of CRE loans with $680M of reserves (FY2025). Transitional/floating-rate CRE lenders meeting the maturity wall at higher refi rates.
• The maturity wall: ~$875B due 2026 (MBA, 17% of balances) vs ~$1.148T (S&P) — cite both, methodologies differ; >$4T total 2025-2029, refinancing from 3-4% into 6-7%. Listed-REIT equity market cap >$1.4T; REITs own >$4.5T of CRE.

Synthesis

mREITs concentrate leverage on the exact agency MBS banks and the Fed hold — so a rate shock forces a levered repo unwind into the market that marks bank balance sheets (the SVB-style channel and the 2020 fire-sale channel are the same pipe). The same long-rate common factor (macro-bank-htm-marks, macro-cross-sectional-analysis) hits mREIT book value, bank securities, and the GSE/Fed MBS book at once; commercial mREITs add the CRE-credit leg into the 2026-27 wall.

What is NOT asserted

• No claim a major mREIT is currently failing — the point is the levered fire-sale mechanism + shared collateral, with 2020/2022 as the realized precedents.
• Maturity-wall figures are contested (MBA ~$875B vs S&P ~$1.148T for 2026) — both cited.
• March-2020 peer specifics beyond Invesco Mortgage are contested.
• Overlay edges are excluded from the proofs.

Sources: Annaly Q1 2026 8-K; AGNC Q3 2025; AGNC 2022 annual report; Invesco Mortgage 8-K (Mar 2020); Invesco Mortgage 10-Q Q1 2020; BXMT $0.47 dividend; Nareit fact sheet (May 2025); MBA — 2026 maturities; S&P — CRE maturity wall; Urban Institute — April 2025 chartbook; FSOC 2024 annual report.

← Research index · structured data: macro-mortgage-reits.json · macro-mortgage-reits.md