US banks hold about $3.1 trillion in commercial real estate loans, and the official books say almost nothing is wrong. Reported delinquency rates sit below 1.25%, a picture so clean it looks like 2008 never happened. The market that is forced to report everything tells a different story: in September 2026 the CMBS delinquency rate hit 8.02%, its highest since November 2020, and the special servicing rate hit 11.49%, a 13-year high.
The gap between those two pictures has a name, and the New York Fed itself documented it: extend and pretend. Banks are pushing maturities out and trimming payments instead of recognising losses, and the accounting rules let them do it. The question now is not whether the stress is real. It is whether the strategy can hold until refinancing gets cheaper, or whether the maturity wall coming in 2026 and 2027 forces the reckoning.
The Numbers at a Glance
- Bank commercial real estate loans: about $3.14 trillion in September 2026, according to the Federal Reserve H.8 release of 23 October 2026, out of roughly $6.3 trillion in total US commercial real estate debt.
- Who holds it: banks outside the 25 largest hold about 70% of bank commercial real estate loans, and small and mid-sized banks hold roughly two thirds to three quarters of it, according to the same H.8 data and a Wall Street Journal analysis.
- CMBS delinquency, September 2026: 8.02%, up 17 basis points, the highest since November 2020, according to Trepp.
- CMBS special servicing, September 2026: 11.49%, up 7 basis points, the highest since February 2013, on $69.53 billion of loans.
- CMBS delinquency, October 2026: 7.87%, down 15 basis points, but only because a single $975 million showroom loan was cured. Office, multifamily and retail delinquency all rose, and special servicing climbed to 11.74%, the highest since December 2012.
- Office vacancy: 19.8% nationally in the third quarter of 2026, down from a 20.1% peak a year earlier, according to Cushman and Wakefield.
- Maturities: $875 billion of commercial and multifamily mortgages were scheduled to mature in 2026, with $396 billion of that at banks, and a further $652 billion across all lenders in 2027, according to the Mortgage Bankers Association.
What Extend and Pretend Actually Means
A commercial mortgage usually runs about five years, with the whole balance due at once at the end. When that maturity arrives and the building is worth less than the loan, or the rent no longer covers a refinancing at today’s rates, the borrower cannot pay it off. The bank has two choices: force a sale, recognise the loss and take the hit to its capital, or extend the loan, trim the payment and wait.
Extend and pretend is the second choice, repeated at scale. The New York Fed documented it in Staff Report 1130, by Matteo Crosignani and Saketh Prazad, first published in October 2024 and revised in June 2026: weakly capitalised banks extended distressed commercial loans and granted payment relief to preserve their capital, and the extended loans could still count as performing. At the end of 2023, commercial mortgages coming due within three years equalled 27% of bank capital.
The accounting rules make the strategy possible. Under ASU 2022-02, banks must report a loan modification only when they determine the borrower is experiencing financial difficulty. The banks themselves make that call. An underwater borrower can receive a balloon-payment extension classified as a routine renewal, and no troubled-debt flag ever appears.
The scale of the omission is visible in the official numbers. A year ago, the St. Louis Fed showed banks had officially reported just $27.7 billion in modified commercial real estate loans for borrowers in financial difficulty, less than 1% of a market worth more than $3 trillion. Yet Federal Reserve Board research by David Glancy, published as FEDS 2026-025 in May 2026, found that large banks actually extended roughly half of all maturing commercial real estate loans between 2023 and 2025. The gap between $27.7 billion reported and roughly half of all maturities extended is the size of the pretence.
There is a counterpoint, and it matters. Glancy’s paper also found that 58% of stress-era extensions paid off within six quarters, which suggests many extensions are bridges to a genuine recovery rather than pure delay. But that study covers only banks above $100 billion in assets. The banks carrying most of the exposure are far smaller.
Two Delinquency Rates, Two Different Stories
The cleanest way to see the gap is to put the two delinquency measures side by side. Bank books report commercial real estate delinquency of around 0.73% to 1.25%, according to FDIC Call Report filings. The securitised market, where trusts are legally forced to disclose distress, reports something very different.
Trepp’s September 2026 data, published in October, put the overall CMBS delinquency rate at 8.02%, up 17 basis points from August and the highest since November 2020. The special servicing rate, which tracks loans transferred to a special servicer for default, imminent default or other distress, rose 7 basis points to 11.49%, its highest since February 2013. New transfers totalled $2.65 billion across 42 loans in the month, 57% of it office.
October’s data, published in early November, looks better at first glance and worse underneath. The headline delinquency rate fell 15 basis points to 7.87%, but Trepp attributed the drop to a single $975 million showroom-property loan returning to current status. Every major property type got worse at the same time.
| Property type | Delinquency, September 2026 | Delinquency, October 2026 |
|---|---|---|
| Overall CMBS | 8.02% | 7.87% |
| Office | 12.16% | 12.33% |
| Multifamily | 8.04% | 8.09% |
| Retail | 6.58% | 7.21% |
| Lodging | 6.18% | Not yet reported |
| Industrial | 1.14% | Not yet reported |
| Special servicing rate | 11.49% | 11.74% |
Office remains the most distressed type, at 12.33% in October, and the special servicing rate rose another 25 basis points to 11.74%, the highest since December 2012. The two measures disagree because they measure different things: delinquency counts loans already past due, while special servicing catches loans heading for trouble before they miss a payment. Both are now at or near their worst readings since the aftermath of the financial crisis.
Where the $3 Trillion Sits
The exposure is not spread evenly across the banking system, and that is what makes it a systemic question rather than a collection of bad loans. Smaller regional and community lenders, a sub-sector that commands only about 13.5% of industry assets, carry nearly two thirds of the commercial real estate exposure.
Research from Columbia, Wharton and MIT, published as an NBER working paper in January 2026, found that 53% of US regional banks already exceed the 300% commercial real estate to capital threshold, with a median concentration of 310%. The 2006 interagency guidance from the Federal Reserve, the FDIC and the OCC says total commercial real estate loans at 300% or more of capital, and construction loans at 100% or more, trigger heightened supervisory scrutiny. The FDIC counted 1,374 banks, roughly 31% of all banks, as commercial real estate concentrated at the end of 2024.
Commercial mortgages represent about 38% of total assets at a typical regional bank, against about 8% at the money-centre banks. That concentration is why a problem in office buildings becomes a problem for bank capital, and why the official delinquency numbers look so implausible next to the CMBS data. A bank with 38% of its balance sheet in commercial property has every incentive to extend a troubled loan rather than crystallise the loss.
The Maturity Wall in 2026 and 2027
Extend and pretend buys time, but time only helps if refinancing gets cheaper before the next maturity arrives. The calendar is not cooperating. The Mortgage Bankers Association’s February 2026 survey found $875 billion of commercial and multifamily mortgages scheduled to mature in 2026, 17% of the $5.0 trillion outstanding, with $396 billion of that sitting at banks. A further $652 billion matures across all lenders in 2027.
Borrowers refinancing today face roughly double the cost of the loans they took out in the low-rate years. The Federal Reserve held its benchmark rate at 3.75% to 4.00% on 28 October 2026, and commercial loans that were written at near-zero rates now have to be refinanced at far higher ones. New commercial mortgage lending has fallen 4.8% to 5.3% since early 2022, as old loans crowd out new credit.
The office market itself is stabilising, but slowly and selectively. National vacancy stood at 19.8% at the end of the third quarter of 2026, according to Cushman and Wakefield, down from a 20.1% peak in the third quarter of 2025. Net absorption was positive for the fifth straight quarter, at just 50,000 square feet in the quarter and 15.5 million square feet over the past year. Class A vacancy fell 60 basis points year over year, and top-tier buildings captured two thirds of third-quarter leasing. Analyst estimates reported by the Financial Times put office values down roughly 40% to 50% from peak, which is why even a stabilising vacancy rate does not fix the refinancing maths for older buildings.
What the Banks Themselves Just Reported
Third-quarter earnings from the regional banks show why the bear case is not straightforward. Western Alliance Bancorporation reported earnings per share of $2.45, up 7.5% year over year, with net income of $277.2 million. Nonaccrual loans fell to 0.79% of funded loans, and the bank lifted its allowance coverage of nonaccrual loans to 118%, up from 96% in the second quarter. Zions Bancorporation reported earnings per share of $1.78, up 20% year over year, with net earnings of $262 million and net loan charge-offs of just $3 million, down from $56 million a year earlier.
Those results are the strongest argument for the banks: actual credit losses remain small, earnings are growing, and reserves are being built. The counterargument is that reported losses are small precisely because extend and pretend keeps troubled loans off the loss column. Both readings use the same numbers.
Why Shorting Regional Banks Is a Harder Trade Than It Looks
The bear thesis, as set out in the Benzinga analysis that prompted this article, treats the public CMBS market as a mirror for regional bank loan books. That mirror is distorted in three ways.
First, securitised trusts disproportionately hold the highest-leverage, non-recourse transitional debt, the riskiest slice of the market. Bank-held multifamily delinquencies were as low as 1.4% in the second quarter, and Freddie Mac’s $500 billion-plus apartment portfolio remains below 0.8% delinquent. The worst loans live in CMBS; the best live on bank balance sheets.
Second, shorting a whole sector means shorting the strong with the weak. Broad bets against vehicles like the SPDR S&P Regional Banking ETF (KRE) and the iShares US Regional Banks ETF (IAT) carry exposure to banks that just reported growing earnings and rising reserve coverage, not only to the most concentrated lenders.
Third, consolidation caps the downside for individual names. The number of US banks has fallen from about 18,000 in 1986 to just over 4,200, according to Marketplace, and the trend is still running. A distressed bank with a valuable deposit franchise or a strong regional position can be acquired at a premium, which is the worst outcome for a short seller.
None of that means the stress is imaginary. The CMBS data, the Fed’s own research and the concentration figures all point the same way. It means the trade is not the one-way bet the headline delinquency rate suggests, and the banks with the cleanest reported books are not necessarily the safest.
What to Watch Next
Four data points will decide whether extend and pretend ends in a soft landing or a reckoning. Trepp’s November CMBS readings will show whether October’s improvement in the headline rate survives without another large loan cure, and whether special servicing keeps climbing past 11.74%. Fourth-quarter bank earnings in January will show whether charge-offs stay near zero or start to rise as 2026 maturities are worked through. The Federal Reserve’s December meeting will shape borrowing costs for owners heading into the heavy 2027 maturity calendar. And the Federal Reserve’s Financial Stability Report will show whether regulators are still describing commercial real estate as a contained risk, or starting to use stronger language.
The $3 trillion is not going anywhere quickly. The only question is whether it is worked out loan by loan, at a pace earnings can absorb, or whether the maturity wall forces a faster answer.
Frequently Asked Questions
What is extend and pretend in commercial real estate?
Extend and pretend is the practice of extending a maturing commercial mortgage and trimming its payments instead of forcing a sale and recognising a loss. The New York Fed documented it in Staff Report 1130, finding that weakly capitalised banks extended distressed commercial loans to preserve their capital, and the extended loans could still count as performing.
How much commercial real estate debt do US banks hold?
US banks held about $3.14 trillion in commercial real estate loans in September 2026, according to the Federal Reserve’s H.8 release, out of roughly $6.3 trillion in total US commercial real estate debt. Banks outside the 25 largest hold about 70% of the bank total.
What is the CMBS delinquency rate now?
Trepp put the overall CMBS delinquency rate at 7.87% in October 2026, down from 8.02% in September, when it was the highest since November 2020. The October fall came from a single $975 million loan cure, while office, multifamily and retail delinquency all rose. The special servicing rate reached 11.74% in October, the highest since December 2012.
Why do bank delinquency rates look so much lower than CMBS rates?
Bank books report commercial real estate delinquency of around 0.73% to 1.25%, while the CMBS market reports 7.87%. The gap comes from discretionary accounting: under ASU 2022-02, banks report a modification only when they judge the borrower to be in financial difficulty, and they make that judgement themselves. CMBS trusts are legally forced to disclose distress.
How many regional banks are over the commercial real estate concentration limit?
Research from Columbia, Wharton and MIT found that 53% of US regional banks exceed the 300% commercial real estate to capital threshold, with a median concentration of 310%. The FDIC counted 1,374 banks, about 31% of all banks, as commercial real estate concentrated at the end of 2024.
Is shorting regional banks a good trade?
It is riskier than the CMBS numbers suggest. CMBS trusts hold the riskiest slice of the market, bank-held multifamily loans are performing far better, broad short vehicles include strong banks reporting earnings growth, and consolidation means distressed banks can be bought at a premium. The stress is real, but the trade is not one-sided.
About the Author

Andy Abramson
Andy Abramson writes about luxury real estate, waterfront communities, and high-end lifestyle destinations. Andy’s guides focus on the detail behind a property, not just the listing: property records by address, name, and folio, closing timelines by loan type and state, and deeds and transfer rules checked against official sources, public records, and major news outlets. Andy also fact-checks high-profile homes and luxury rankings before publishing.













