You extended twelve months ago. The memo said what extensions always say — a term, a fee, a replacement cap, a new maturity date. What it did not say, anywhere in writing, was the rate you were going to refinance at.
That number existed. It just lived in the decision rather than the document. You extended instead of refinancing because refinancing then looked expensive relative to refinancing later, and "later" meant a specific place you expected the index to be. Nobody wrote it down, so nobody has gone back to check it.
Go check it. It is the only assumption in the file that has moved against you, and it is the one nobody is tracking.
What changed between the extension and today
We said in June that the cut had come off the table. This is worse than that, and the difference is not a matter of degree.
June's argument was that the easing everyone had penciled in was not arriving — that the path had flattened rather than fallen. The sign has since flipped. The market is no longer pricing a delayed cut. It is pricing tightening: roughly 24 basis points of it by the end of 2026, with little change through 2027. Three-month term SOFR sits at 374 basis points. The Fed held at 3.50%–3.75% on July 29 for the fifth consecutive meeting, and July's CPI — 0.1% on the month, 3.4% annual, 2.5% core, released August 12 — was benign enough that the market read it as a reason to keep holding rather than a reason to cut. Odds on a cut at the September meeting are effectively zero; the live debate is hold versus hike.
"Cut delayed" and "hike priced" are different claims, and only the second one invalidates an extension. A delayed cut costs you carry while you wait. A priced hike means the thing you were waiting for is no longer on the schedule in either direction, and the reason you chose to wait has been removed rather than postponed.
The long end offers no compensation. The 10-year Treasury was 4.63% and the 2s10s spread 48 basis points as of August 13. A curve that flat does not pay you to term out — the extra duration you take on in a fixed-rate takeout buys you no discount. This is the trap in reverse: sponsors who understand that their carry prices off the short end often assume a flat curve helps the refinance. It does not. It just removes the reward for having waited.
Keep the two ends separate when you do the math, because the temptation is to net them. The 24 basis points of tightening the market prices lands on the short end, where it costs a $3,500,000 interest-only bridge about $8,400 a year in carry — real money, but carry, not coverage. Your takeout prices off the long end. A short-end move is not a credit against a long-end rate, and a sponsor who nets one against the other will conclude the exit repriced by less than it did.
The arithmetic of a wrong-direction index
Same file as the last two posts in this series — one operator, one fact base. Value-add multifamily, Northeast secondary market, business plan executed, stabilized. Everything below traces from two numbers: a $3,500,000 maturing balance and $315,000 of stabilized annual NOI.
Three exit rates matter here, and only two of them were ever written down.
| Exit rate | Miss vs. reality | Mortgage constant | |
|---|---|---|---|
| The hope at extension | sub-6% | 65+ bp | 7.1561% at 5.95% |
| The paper — what files underwrote | 6.50% | 12.5 bp | 7.5848% |
| Reality — today's sheet | 6.625% | — | 7.6837% |
The 6.50% and 6.625% figures come from actual extension memos and a current sheet. The sub-6% "hope" is a characterization of what sponsors said they were waiting for, not a rate anyone underwrote and not market data. It is anchored at 5.95% below purely to make the arithmetic traceable.
Note what amortization does before you read the coverage numbers. Going from 6.50% to 6.625% is 12.5 basis points of rate but only 9.9 basis points of constant. Going from 5.95% to 6.625% is 67.5 basis points of rate but 52.8 basis points of constant. As we showed on this same file in July, proceeds and coverage are set by the constant, not the coupon, and on a 30-year amortization the constant always moves less. That is the reassuring direction, and it is why the answer below is not a catastrophe for everyone.
Now the coverage, at a 1.25x takeout test and at the 1.22x a thinner file was underwritten to:
| Underwritten at | Against | Exits at | vs. 1.20x floor |
|---|---|---|---|
| 1.25x | 6.50% — the paper | 1.234x | clears by 0.034 |
| 1.22x | 6.50% — the paper | 1.204x | on the line |
| 1.25x | sub-6% — the hope | 1.164x | below by 0.036 |
| 1.22x | sub-6% — the hope | 1.136x | below by 0.064 |
The finding, and it is narrower and more useful than the headline suggests: the files underwritten to the paper survive. The files extended on the hope do not.
A sponsor whose memo said 6.50% missed by twelve and a half basis points and is fine. Their coverage went from 1.25x to 1.234x, their carry rose $4,375 a year — $365 a month — and no lender will notice. That is what underwriting to a sheet instead of a forecast buys you, and it is worth saying plainly because the alarmist version of this article would not.
The exposed cohort is the one that extended because relief was assumed, and whose file was thin to begin with. At 1.22x underwritten against a sub-6% exit, coverage lands at 1.136x — below every refinance floor we place against, with no change whatsoever in the property's performance. Same rent roll, same NOI, same executed business plan. The building did nothing wrong. The assumption did.
And the proceeds gap has widened accordingly. At a 1.25x test and today's 6.625%, supportable proceeds are $3,279,656 against a $3,500,000 payoff — $220,344 short. When we sized this same file in July at a 6.25% takeout, the gap was $89,343. Three quarters of a point of takeout rate has more than doubled the equity check.
So what cushion would actually have been enough? Worth knowing, because it is the number to underwrite the next one to. From a 6.625% takeout, a full 100 basis points of further rate moves the constant 81 basis points — 7.6837% to 8.4935% — and costs about 9.5% of your coverage ratio. Run that against the floor: a 1.25x file lands at 1.131x, and even a 1.30x file lands at 1.176x and is still short. You need roughly 1.326x underwritten today to absorb a full point and still clear 1.20x. That is a wider cushion than most 2024 and 2025 value-add files were built with, which is the honest reason this cycle is catching people who did nothing operationally wrong.
The tape
July's CMBS data is worth reading precisely, because the imprecise version of it will be quoted at you.
Trepp's July report put the all-property delinquency rate at 7.86%, up 51 basis points. Multifamily posted the largest property-type increase, up 46 basis points to 7.69%. That is the number that will get repeated. Here is the mechanism behind it: the multifamily rise came from loans going 30 days delinquent — an operating signal, cash flow arriving late — not from borrowers failing to refinance at maturity.
The maturity story sits in a different column. Roughly 66% of newly delinquent balances were loans past maturity, and that concentration is in the all-property figure, led by office and retail rather than apartments.
So: multifamily maturity stress is the forward risk this piece is about. It is not yet what the print is showing. Anyone telling you July's multifamily number proves the refinancing wall has arrived is reading two mechanisms as one.
The counter-case
Twenty-four basis points priced is a lean, not a landing.
Two cuts priced out is not two hikes locked in, and a market that assigns real probability to a September hike is not the same as a Fed that has decided to deliver one. If the funds rate simply sits at 3.50%–3.75% through 2027 — the single most likely path on today's pricing — that is survivable for any well-covered asset. Nothing in the table above breaks a 1.30x file. The sponsor genuinely exposed here is specific: thin coverage, extended on assumed relief. If that is not you, the correct response to this article is to verify your number and move on.
There is a second, larger caveat, and it is the reason this piece is dated rather than timeless. Kevin Warsh delivers his first Jackson Hole keynote as Fed chair at the symposium running August 27–29. A new chair's first symposium address is a genuine two-sided event. If it lands dovish, some of the repricing described above partially reverses and the extension math looks less wrong than it does today. If it lands hawkish, the 24 basis points currently priced is a floor rather than a lean.
We are writing this on August 17, 2026, ten days ahead of that. That is deliberate: the argument here is a pre-registered position, not a reaction. It stands or falls partly on what Warsh signals, and saying so before the fact is the only version of this claim worth making. We will report against it when the Q2 tracker data lands in September, the same commitment we made in the SLOOS post.
The sixty days
Reprice the exit at the current forward index, not the one in your extension memo. This is a twenty-minute exercise and almost nobody has done it. Pull the rate your extension assumed — if it is not in the memo, it is in whoever's model justified the decision — and put today's sheet next to it. The gap is your exposure.
Then get your actual coverage number. Not the one from the extension package. The one that falls out of today's constant. If it starts with 1.1, you have a cash-in refinance to plan for and roughly sixty days is not a lot of time to plan it in.
From there the playbook is the standing one, and it has not changed: arrive pre-underwritten, have the exit identified before you need it, and know where your incumbent bank sits on the CRE concentration table before the renewal conversation starts. On that last point there is a genuine tailwind — the banks that eased multifamily credit last quarter were exclusively the ones above $100 billion in assets, and the cohort that holds most New Jersey CRE reported weaker loan demand, which means the queue at the institutions that matter for a $1M–$5M file is currently short. A thin queue is worth more than a quarter point when your maturity is fixed. The NJ Bank CRE Lending Tracker has the bank-by-bank position, and July's capacity analysis explains why concentration, not appetite, is the binding constraint at most of them.
One line to carry out of this: a coverage ratio that cleared underwriting can now sit below the floor with nothing at all having happened to the property. Whether that describes your file depends entirely on whether you extended against a sheet or against a hope.
FAQ
What exit rate should a 2026 bridge extension be underwritten to? The rate on a current sheet, not the rate on the forward curve you extended against. On the file we have traced through this series — a $3,500,000 balance against $315,000 of stabilized NOI — extension memos written in late 2025 and early 2026 underwrote a 6.50% exit. Today's small-balance multifamily takeout is 6.625%. That is a 12.5 basis point miss, and on a 30-year amortization it moves the mortgage constant only 9.9 basis points, from 7.5848% to 7.6837%. A file underwritten to 1.25x coverage against that 6.50% assumption exits at 1.234x and still clears a 1.20x floor. The files in trouble are not the ones that underwrote conservatively and missed by a little. They are the ones that underwrote the hope.
Does a flat yield curve help a floating-to-fixed refinance? No, and this is the part sponsors most often have backwards. The two rates in a bridge-to-permanent trade sit at opposite ends of the curve: your floating carry prices off the short end, and your takeout prices off the long end. A flat curve — the 2s10s spread was 48 basis points as of August 13, 2026 — means terming out at the long end buys you no discount for the duration you are taking on. It does not lower your exit rate; it removes the reward for waiting. Meanwhile the roughly 24 basis points of Fed tightening the market now prices by year-end lands on the short end, where it costs a $3,500,000 interest-only bridge about $8,400 a year in carry. Do not credit a short-end move against a long-end rate. They are different rates and only one of them sizes your loan.
Is rising multifamily CMBS delinquency a performance problem or a maturity problem? In July's data it is a performance signal, and the maturity stress is the separate forward risk. Trepp's July 2026 report put the all-property delinquency rate at 7.86%, up 51 basis points, with multifamily posting the largest property-type increase at 7.69%, up 46 basis points. That multifamily rise was driven by loans going 30 days delinquent — an operating and cash-flow signal, not borrowers failing to refinance at maturity. The matured-balloon concentration sits in the all-property figure, where roughly 66% of newly delinquent balances were loans past maturity, led by office and retail rather than apartments. Keep the two mechanisms apart. Conflating them turns a coverage problem into a refinancing problem a quarter before the data supports it.
How much coverage cushion absorbs a 100 basis point move in the exit index? About 9.5% of your coverage ratio, which means you need roughly 1.33x underwritten today to still clear a 1.20x floor after a full point. From a 6.625% takeout, 100 basis points of rate moves the mortgage constant 81 basis points, 7.6837% to 8.4935%, and multiplies coverage by 0.905. A 1.25x file lands at 1.131x. A 1.30x file lands at 1.176x — still short. Only at about 1.326x underwritten does a 100 basis point miss leave you at the floor. That is the honest cushion, and it is wider than most 2024 and 2025 value-add files were built with.
Send Us the Extension Memo and the Current Sheet
If you extended a floating-rate bridge in the last eighteen months, the useful exercise takes an afternoon and you can start it without us: find the exit rate that decision assumed, and put a current sheet beside it. If the gap is inside about forty basis points, your file almost certainly still clears and you can stop reading. If it is wider, or if you were underwritten thin to begin with, the number worth having is your coverage at today's constant — not the one in the extension package.
Send us the balance, the NOI, the amortization the takeout will carry, and what the extension assumed, and we will reprice the exit against current sheets, tell you which test binds, and run the refinance against another extension so you are comparing two decisions rather than defaulting into one. We arrange across 100-plus bank, agency, and private-credit relationships, which is the part that decides whether a 1.13x file has a home. If the answer is that you are fine and should sit still, that is a good afternoon too.
Dominick Prevete — Founder, Blue Sky Capital Advisors. 31 years in real estate finance; 100+ bank and private-lender relationships across the bridge, agency, and private-credit spectrum. 4 Sutton Ct, Hamburg, NJ 07419 · (908) 220-6404. NMLS information available upon request. Lending in all 50 states.
The $3,500,000 balance, $315,000 NOI, and 30-year amortization are carried forward from our July 30 and August 5, 2026 analyses on the same file. The 6.50% underwritten exit rate is from extension memos written in late 2025 and early 2026; the 6.625% current takeout is from a lender sheet as of August 2026 — both supplied by the author and neither is a quote. The sub-6% "hope" is a characterization of sponsor expectation at extension, not market data, and is anchored at 5.95% solely to make the arithmetic traceable. Mortgage constants are computed on a 30-year amortization; coverage ratios scale by the constant, not the coupon. Term SOFR, the 10-year Treasury, the 2s10s spread, forward-priced Fed tightening, and September FOMC probabilities are as of August 12–14, 2026 and move daily — re-check before relying on any of them. CPI figures are for July 2026, released August 12, 2026. Delinquency figures are from Trepp's July 2026 CMBS report. Jackson Hole symposium dates are the announced dates; keynote timing by the chair is convention, not a published agenda. Every deal prices to its own sheet on its own day. This is market commentary, not a commitment to arrange financing, an offer of terms, or investment advice.
Loans are for business purposes only and are not subject to TILA, RESPA, or HOEPA. Not for primary residences. Equal Housing Opportunity. All loans subject to underwriting approval. Rates and terms shown for illustration; actual rates depend on deal specifics. We do not lend to borrowers with credit below 600 or on owner-occupied properties.