In 2025 alone, roughly $957 billion of US commercial real estate debt matured — a record, per the Mortgage Bankers Association — most of it originated in 2020-21 at rates that no longer exist.[1] A June 2023 interagency policy statement from the Fed, FDIC, OCC and NCUA explicitly permits banks to modify or extend those loans without automatically reclassifying them as distressed — the regulatory license behind what the industry calls “extend and pretend.”[2] A 2024 NY Fed working paper found weakly capitalized banks disproportionately used exactly this option to avoid recognizing losses, crowding out new lending in the process.[3] The strain concentrates in one property type and one class of lender: office, where CMBS delinquency hit a record ~10-11% and values fell roughly 40% from the 2022 peak, sits on the books of regional and community banks, which hold roughly two-thirds of all bank-held CRE debt.[4][5] New York Community Bancorp's January 2024 surprise loss remains the marquee event — but the honest, load-bearing caveat is that office is only about 15-20% of the total CRE debt universe, no second CRE-driven bank failure has been confirmed since, and the Fed's own stress tests show large banks absorbing a hypothetical 40% CRE decline while staying above capital minimums.
Most of the commercial real estate loans maturing right now were underwritten between 2020 and 2021, when financing rates sat near zero and cap rates were compressed to match. Roughly $957 billion of that debt matured in 2025 alone — a record, and the Mortgage Bankers Association's own survey number, before counting the loans that got rolled into 2026-27 by extension rather than resolved.[1] Refinance one of those loans today, at today's rates and today's cap rates, and the math often doesn't clear: the new loan proceeds fall short of the old balance, and somebody has to either write a check or write down the value.
There's a third option, and it's the one that's actually been used at scale: don't refinance, extend. A June 2023 interagency policy statement from the Fed, FDIC, OCC, and NCUA explicitly permits banks to modify or extend a maturing CRE loan without automatically classifying it as distressed.[2] A 2024 NY Fed working paper, circulated via Liberty Street Economics, found that weakly capitalized banks used this option disproportionately — not to buy time for a genuine recovery, but to avoid recognizing a loss they had reason to believe was already there, and that the practice crowded out new CRE lending as a side effect.[3] The appraisal on the books stays where the extension says it should be. The transaction that would prove otherwise simply doesn't happen.
The strain isn't evenly spread — it's concentrated almost entirely in one property type. Office vacancy sits near a record ~20%; office values are down roughly 40% from their 2022 peak; office CMBS delinquency hit a record ~10-11% through 2025, the worst-performing major property type by a wide margin, while overall CRE delinquency at banks stayed in the low single digits.[4] And it's concentrated in one class of lender: small and regional banks hold roughly two-thirds of all bank-held CRE debt, versus a small single-digit share at the money-center banks, which is why New York Community Bancorp's January 31, 2024 surprise loss — a ~$552M provision, a ~70% dividend cut, a stock that fell 38% in a day, and eventually a $1.05B rescue led by Steven Mnuchin's Liberty Strategic Capital — remains the reference event.[5]
The honest caveat, and it's a real one: office is only about 15-20% of the total CRE debt universe — multifamily, industrial, and retail are materially healthier — and no second CRE-driven bank failure has been confirmed since NYCB. The Fed's own DFAST stress tests, which model a hypothetical ~40% CRE price decline, show large banks staying above capital minimums through the scenario. This is a slow, known, largely-reserved-against grind, not a surprise run. Whether extend-and-pretend is buying time for a genuine recovery or simply deferring a larger reckoning is the open question this diagnostic states rather than resolves — and it's the same structural question UC-256 asks of private credit.
MBA's own survey figure for 2025 maturities; cumulative 2024-27 figures range $2.0-2.8T depending on whether rolled extensions are counted — a disagreement that is itself evidence of the mechanism.[1]
How a regulatory policy became a market-wide practice, and where it broke first.
CRE loans underwritten at near-zero rates and compressed cap rates. The financing environment that made these loans work no longer exists by the time they mature.
The VintageThe Fed, FDIC, OCC and NCUA jointly issue a policy statement permitting banks to modify or extend maturing CRE loans without automatic adverse classification — the regulatory license behind extend-and-pretend.[2]
The LicenseNew York Community Bancorp reports a surprise Q4 loss, cuts its dividend ~70%, and books a ~$552M provision driven substantially by office/CRE exposure. Stock falls 38% in a day.[5]
The Marquee EventA ~$1.05B equity injection led by Steven Mnuchin's Liberty Strategic Capital stabilizes NYCB; the bank later rebrands as Flagstar Financial. The precedent for how a CRE-driven bank crisis gets contained.[5]
The RescueExtend-and-pretend allowed weakly capitalized banks to avoid recognizing losses — and crowded out new lending in the process. — NY Fed / Liberty Street Economics working paper, 2024
| Dimension | Evidence |
|---|---|
| Quality (D5) Origin · 88 | The lever is whether a carrying value reflects reality: the 2023 policy statement lets banks extend a maturing loan without marking the loss the extension is implicitly conceding.[2][3] D5 is the origin because this case is fundamentally not about whether CRE has problems — it's about whether the number on a bank's books is allowed to say so before a transaction forces it.The Honesty of a Number |
| Operational (D6) L1 · 84 | The operational fact underneath the mark-honesty question: $957B matured in 2025 alone, a record, most underwritten at rates and cap rates that no longer exist.[1] D6 amplifies from D5 because the wall is what forces the question — every maturity is a moment where extend-and-pretend either continues or a real number gets printed.The Maturity Wall |
| Revenue (D2) L1 · 78 | The revenue/capital exposure concentrates in regional and community banks, which hold roughly two-thirds of all bank-held CRE debt — a far larger share of their balance sheets and capital than at money-center banks.[5] D2 amplifies alongside D6: the same maturity wall is a rounding error for a diversified money-center bank and a capital event for a concentrated regional one. |
| Customer (D1) L2 · 70 | NYCB tested the depositor/investor confidence question directly: a single earnings surprise triggered a 38% one-day stock decline and a scramble for outside capital.[5] D1 sits here because it demonstrated the actual transmission mechanism — not a regulator forcing a bank to recognize a loss, but the market doing it the moment a number looked wrong. |
| Regulatory (D4) L2 · 72 | D4 is not a bystander dimension here — the 2023 interagency policy statement is the regulatory act that enables the entire mechanism this case documents.[2] It sits at l2 rather than origin because the policy is permissive, not causal: it allows extend-and-pretend, it doesn't require it, and DFAST stress tests suggest the regulatory system also has a countervailing capital-adequacy check running in parallel. |
| Employee (D3) 48 | Deliberately the thinnest dimension in this case. Unlike UC-256's fraud prosecutions reaching named individuals, the CRE mark-to-fantasy mechanism is a balance-sheet and regulatory story, not a workforce one — no comparable employee-level cascade has surfaced in the research for this case. |
The cascade originates in D5 — Quality — because the lever is fundamentally about the honesty of a number: whether a bank's carrying value for a loan reflects what the collateral could actually sell for today.[2][3] From D5 it amplifies into D6 (the operational reality — a record $957B maturity wall, concentrated in a distressed property type) and D2 (the regional-bank revenue and capital exposure, since two-thirds of bank-held CRE debt sits with small institutions).[1][4] It then reaches D1 (the depositor and investor confidence question NYCB tested directly) and D4 (the regulatory dimension — the 2023 policy statement is itself the enabling mechanism, not a bystander).[5][2] D3 is the thinnest dimension here — this is a balance-sheet cascade, not a workforce one. Cross-references: [UC-039] is the deposit-run precedent regional banks are still capitalized against; [UC-256] runs the identical stale-mark mechanism in private credit — different asset, same physics; [UC-259] is the counter-cascade this case must respect — buffers and slow recognition may mean contained, not systemic, and that argument gets made in full there.
-- UC-257: Mark-to-Fantasy: 6D Diagnostic Cascade
-- $957B CRE maturity wall, extend-and-pretend enabled by 2023 policy (cluster: UC-256/258/259/260; counter: UC-259)
FORAGE mark_to_fantasy
WHERE appraisal_exceeds_transaction_reality = true
AND extend_and_pretend_enabled_by_policy = true
AND office_concentrated_in_regional_banks = true
ACROSS D5, D6, D2, D1, D4, D3
DEPTH 3
SURFACE mark_to_fantasy
DIVE INTO extend_and_pretend_mechanism
WHEN refinancing_forces_recognition = true
AND loan_extended_instead = true
TRACE mark_to_fantasy_cascade
EMIT cre_stale_mark_signal
DRIFT mark_to_fantasy
METHODOLOGY 90
PERFORMANCE 48
FETCH mark_to_fantasy
THRESHOLD 1000
ON MONITOR CHIRP high '$957B in CRE debt matured in 2025 alone, underwritten at rates that no longer exist. A 2023 interagency policy statement lets banks extend rather than mark the loss. Office (only ~15-20% of CRE) carries record ~10-11% CMBS delinquency and ~40% value declines, concentrated in regional banks holding two-thirds of bank CRE debt. NYCB remains the marquee event; no second CRE bank failure confirmed since'
SURFACE analysis AS jsonRuntime: @stratiqx/cal-runtime · Spec: cal.semanticintent.dev · DOI: 10.5281/zenodo.18905193
The June 2023 interagency policy statement was issued just months after SVB — while regulators were tightening bank transparency with one hand, they were loosening CRE-loss recognition with the other. The mechanism that keeps office losses hidden is regulator-approved, not a loophole.[2]
Nearly every alarming figure in this case — the ~20% vacancy, ~40% value decline, ~10-11% delinquency — is office-specific. Multifamily, industrial, and retail delinquencies remain low. Blurring office with total CRE is the single most common analytical error in this space.[4]
The NY Fed's finding wasn't that all banks use extend-and-pretend — it's that weakly capitalized banks use it disproportionately, meaning the practice concentrates risk exactly where the buffer to absorb a loss is thinnest.[3]
One marquee event in over two years, with no confirmed repeat, is genuinely different from a systemic thread. The DFAST stress tests show large banks absorbing a hypothetical 40% CRE decline. This case names the mechanism; it does not claim the mechanism has failed.
Five sources: the MBA maturity-wall survey, the named 2023 interagency policy statement, the NY Fed academic paper documenting its use, office-vs-CRE delinquency data from Trepp, and the NYCB event with primary company/regulatory filings.
$957 billion matured in 2025 at rates that no longer exist. The regulation that lets banks not find out what it's worth is the real story.