The 10-Year Just Closed at 5.29%, Its Highest in 19 Years. New York Apartments Still Trade at a 5 Cap.

The risk-free rate just passed the cap rate. Nobody in commercial real estate has repriced.

On September 30, the 10-year Treasury closed at 5.29%, according to Treasury's own daily yield curve. That is its highest level in 19 years. It rose 85 basis points in the third quarter alone, more than it moved in the entire first half of the year.

Here is the number that should have ended a lot of acquisition committee meetings last week. In CBRE's H1 2026 Cap Rate Survey, stabilized multifamily in New York City trades at a 5.0% to 5.5% cap rate. The 10-year pays 5.29%. For a meaningful slice of the best-located apartment stock in America, the risk premium for owning real estate instead of a government bond is now approximately zero. In some of it, the premium is negative.

That is not a forecast. That is subtraction, done with two published numbers.

That NYC band was 4.5% to 5.0% in H2 2025, so the survey has already conceded 50 basis points of repricing. It conceded them while the 10-year moved 110. And the same survey says roughly 60% of the people pricing those deals expect cap rates to hold steady for the next six months.

The industry told you its own break-even, then walked past it

The most honest line in CBRE's survey is the one almost nobody quoted. More than 200 CBRE professionals across 50-plus markets were asked what it would take to restart deal flow. The survey reports that "the median respondent thinks the 10-year Treasury would have to fall to 3.75%," well below the 4.6% mark in July.

So the industry's own median dealmaker named a threshold: 3.75%.

The 10-year closed the quarter at 5.29%. That is 154 basis points in the wrong direction from the level the market said it needed. Not 15 basis points, not a rounding error inside a forecast band. A gap as wide as the entire risk premium investors used to demand for owning a building instead of a bond.

Look at the path. On January 2 the 10-year was at 4.19%. On June 30 it was 4.44%. On September 30 it was 5.29%. That is 110 basis points year to date, and 85 of them landed in the third quarter.

The 30-year Treasury closed at 5.64% and the 20-year closed above it at 5.68%, an inversion at the long end that says the bid for duration is thinner than the coupon suggests.

The cause is not mysterious and it is not the Fed. Energy did it. Regular gasoline is up 47% since late February and diesel is up 70%, driven by the wars in Iran and Ukraine and a Russian diesel export ban. KPMG chief economist Diane Swonk's read on the inflation data was blunt: "Cooler on paper, hot underneath." On core PCE she added, "Sticky is an understatement."

Diesel isn't a line item you hedge away. It's in every roof replacement, every turn, every unit of material that gets trucked to your site.

The Numbers: the spread, the subtraction, and the 26% nobody has taken

Haven't cap rates and Treasuries decoupled before, for years at a time, while the people waiting for the spread to normalize missed a decade of appreciation? Yes. They have.

Here is why this time the gap closes from the cap rate side instead.

Start with the spread, because the spread is the whole argument. Historically, core commercial real estate traded at roughly 150 to 250 basis points over the 10-year. That premium paid you for illiquidity, for vacancy, for the capital expenditure you cannot escape, and for the simple fact that a Treasury note does not call you about a water heater.

Run the arithmetic on a 5.0% cap NYC apartment deal against a 5.29% 10-year:

• Current risk premium: negative 29 basis points.
• Cap rate required to restore a 150 bps premium: 6.79%.
• Price change on identical net operating income: down 26%.
• Cap rate required to restore a 200 bps premium: 7.29%.
• Price change on identical net operating income: down 31%.
• At the top of the band, a 5.5% cap, the premium is positive 21 basis points and the required repricing is down 19%. That is the optimistic case.

Nothing in that arithmetic requires rents to fall, vacancy to rise, or a recession to show up.

It requires only that buyers eventually demand the same premium they demanded for decades. CBRE's all-property average sat at roughly 6.6% in H1 2026, which means the pressure isn't confined to trophy multifamily in Manhattan. CRE Daily's read on the same data is that cap rates may still have room to rise.

The debt side closes the trap. Freddie Mac's Primary Mortgage Market Survey put the 30-year fixed at 7.28% on October 1, up from 7.03% the week before and 6.34% a year ago. The 15-year printed 6.60%. Daily lender pricing ran as high as 7.6%, the highest since late 2023. Buy a 5-cap asset with any leverage at that cost and every borrowed dollar reduces your return. That is negative leverage, and the only way it ever worked was as a bet on a refinance the bond market has now priced out of the hold period.

Meanwhile the residential market is already absorbing it. The National Association of Realtors' August existing-home sales report came in at a 3.98 million annual rate, down 2.0% month over month, with inventory up 3.2% to 1.62 million homes. NAR chief economist Lawrence Yun's explanation was one sentence long: "Mortgage rates and home sales move in opposite directions, so it's not surprising to see a mild dip in home buying activity due to high mortgage rates." Houses already repriced, through volume and standing inventory.

Commercial repriced through a survey that says hold steady.

Common Mistakes Investors Make Here

• Underwriting the consensus instead of the curve. If your model assumes a 3.75% or even a 4.5% ten-year inside the hold period, you're not underwriting, you're wishing. Use today's curve, then stress it 75 basis points higher.
• Setting an exit cap equal to or below the going-in cap. On a 5-cap purchase in a 5.29% world, a flat exit cap is the single most expensive assumption in the model. Underwrite your exit 100 to 150 basis points above your entry and see whether the deal survives.
• Calling a bridge loan a business plan. Floating-rate bridge debt priced against a 5.29% long end, with a rate-cut refinance as the exit, isn't value-add. It's a short volatility position with a maturity date.
• Anchoring to the seller's basis. Their 2021 purchase price, their debt and their broker's opinion of value don't matter to what the asset is worth against a 5.29% risk-free rate. Price off the curve, not off their regret.

Where I Land

Cap rates are going up, not sideways, and the roughly 60% of CBRE's respondents calling for no change are going to spend 2027 revising. I'll say it with a date: by June 30, 2027, CBRE's all-property average cap rate prints above 7.0%, against 6.6% in H1 2026. What the consensus gets wrong is treating 2022 through 2026 as a rate spike that reverts. A 20-year trading above the 30-year says the marginal buyer of long duration has gone missing, and that buyer is exactly who a 3.75% ten-year requires.

With my own money, in Q4 2026, I don't buy a stabilized asset under a 7 cap, and I don't sign floating-rate debt whose exit requires a rate cut. I buy two things instead. Paper, from owners who need liquidity more than they need yield. And seller-financed deals at 5.5% to 6%, because on an identical $250,000 balance a 5.5% note costs $291 a month less than a 7.28% bank loan, which is $34,900 over ten years that you create at the closing table instead of hoping for at the exit.

The honest hole in this is the denominator. If rents run 6% and expenses don't follow, a 5-cap today becomes a 6-cap on trailing numbers without a dollar of price movement, and I lose this argument on net operating income rather than on the curve. I don't think New York multifamily gets a 6% rent year in 2027. I've been wrong about rent before.

How to Use PropGPT for This

The repricing won't arrive as a market-wide announcement. It arrives one motivated seller at a time, and the data shows you who they are before the broker does.

"Show me 2-to-4 unit properties in Columbus, OH with more than 120 days on market and at least two price reductions, and include the last sale price and sale date for each."

Days on market plus stacked price cuts is the earliest public signal that a seller has stopped defending their number. Pairing it with the last sale price tells you how much room they have left before they're underwater.

"Find absentee-owned single-family rentals in Phoenix, AZ with a mortgage originated between 2021 and 2022, loan amount over $400,000, and give me owner mailing addresses."

This is the negative-leverage cohort. They bought at compressed caps with cheap debt that now needs replacing at 7% and up. They're the sellers who will actually transact in 2027.

"List free-and-clear duplexes and fourplexes in Memphis, TN owned more than 15 years by out-of-state owners, with mailing addresses."

Free-and-clear, long-held, absentee owners are the seller-finance pool. They have no loan to pay off and no reason to care about a 7.28% bank rate, which is exactly why they'll carry a note at 5.5%. Go in with a real offer at a real price, in writing, on the first contact.

"Run the numbers on a $285,000 duplex in Dayton, OH at $2,450 gross monthly rent. Compare 25% down at a 7.28% bank loan to 10% down on a seller-financed note at 5.5%, and show cash-on-cash for both."

This is the trade of the quarter in one prompt. The bank path costs $1,462 a month on $213,750 borrowed. The seller note costs $1,456 on $256,500 borrowed. Same payment, 15 points less cash out of your pocket, and the difference is entirely the coupon.

"Show me absentee-owned properties in Cape Coral, FL where the estimated open loan balance exceeds 90% of estimated value."

High-LTV absentee owners in a softening market are where price discovery actually happens first. Build the list now, mail it quarterly, and be the only call they get when the refinance quote lands.

The Bottom Line

Two numbers were published in the same week. A government note that pays 5.29% and does nothing, and a New York apartment building that pays 5.0% and needs a roof. One of those prices is wrong, and it isn't the Treasury's.

The next twelve months belong to whoever gets to the seller before the seller accepts the math. That's a data problem, not a conviction problem. Build the list of long-held free-and-clear owners and 2021-vintage borrowers in your market this week, start your conversations at a 7 cap, and let the people underwriting a 3.75% ten-year keep bidding against each other for the deals you don't want.