In June the Zumper National Rent Index turned positive year-over-year for the first time since May 2025, one-bedroom rents up 0.4%. The headline reads like recovery. It isn't.
As Zumper's own CEO put it, the national number is “simply the midpoint between two realities.” Where supply is constrained, rents are rising, San Francisco ran +21.9%. Where it isn't, they are still falling: Houston, Austin, and San Antonio each posted double-digit annual declines.
Yardi Matrix calls multifamily “muddling along,” with flat demand expected through the back half of 2026 and the Sun Belt lagging the Northeast and Midwest. The supply overhang is still the story.
And the rate rescue everyone underwrote is not coming. On June 17, in Kevin Warsh’s first meeting as Fed chair, the Fed held for a fourth straight meeting, stripped forward guidance from a gutted statement, and the median dot now implies a hike, not a cut. The average is a lie, the signal lives beneath it, between markets and, this issue, between property types.
Divergence is the opportunity. Averaging it away is how you miss it. The work is knowing which submarket, and now which sector, sits on the right side of the split.
The underwriting crutch of 2024–2025 was a falling-rate exit. On June 17, a newly hawkish Fed took it away, resetting the math on every floating-rate deal and marginal refinance in the Sun Belt.
Macro only matters when it hits the rent roll, the debt quote, or the expense line. A hawkish Fed hits all three at once, and it ends the strategy of buying soft NOI on faith in a 2026 cut.
Debt is available again. Casual debt is not. This is the difference between liquidity and forgiveness.
Lenders are quoting.
The deal still works after real taxes, real insurance, real concessions, and an exit cap that does not rely on a fantasy Fed path. The market has the first. It does not have the second.
Indicative ranges, informational only, not a commitment to lend. Agency and FHA lead on price; CMBS is widest. The window is open — the deal still has to clear at these coupons.
When acquisition financing crosses roughly 30% of CMBS issuance in a sector, buyers are competing on price, not underwriting to income. Per Trepp, that line has preceded every major CRE correction of the last twenty years.
Office acquisition share has collapsed to 4% in 2025 and 6% so far in 2026; refinancing is 79–84% of originations. On the deals being written today, the signal is quiet.
The danger is the 2021 vintage. Multifamily acquisition share hit 47% that year, underwritten when the 10-year Treasury averaged 1.3%. That paper is hitting refinance now, at 4.4–4.5%.
Two quieter shifts under the headline: the standard loan is getting shorter, and the fastest-growing pocket of lending is the least visible.
The 30% signal is flashing on the refi wall: peak-vintage paper written at 1.3% money, coming due into 4.5% money and a Fed that just deleted its easing bias.
Four of our twelve markets are in Texas, and Texas is where the 2021–23 vintage is meeting its lender. The foreclosure slate has become a monthly print.
This is the 2021-vintage refi wall arriving as actual deeds. The motivated seller here is the lender, the basis resets to today's money, and the buyer who underwrote collected rent at today's money is the one standing across the table. Distress is the entry point, not the warning.
Trepp's price index shows broad stabilization in Q1 2026, but the averages hide it: smaller assets have firmed while the largest properties, and the value-add trade, are still resetting.
Trepp's CRE-CLO data on 1970s-vintage multifamily shows how much of the last cycle's value creation was rate-driven, and how that engine has stalled.
NMHC's June survey: starts are split, 55% of firms unchanged, 22% starting more, 20% fewer. Among those pulling back, the reasons are economic, not logistical.
Source: NMHC Quarterly Construction Survey, June 2026.
12-market multifamily summary, rent growth, vacancy, employment, price per unit, pipeline, net deliveries, and trailing transaction volume. Source: CoStar / Yardi Matrix DataExport, Q1 2026 (as of Q2 2026).
Net-lease volume topped $50B for a third straight quarter, and industrial, not retail, is where it concentrated, with institutions quietly taking share from private buyers.
Source: Multi- & Single-Tenant Market Snapshots, Q1 2026.
Foot-traffic data says the retail story behind the multifamily strip: spending holds, but volumes are softening and the wallet is moving toward value and experience.
Source: Colliers / Placer.ai, May 2026, visits per location, YoY.
Two forces are quietly reshaping where Sun Belt demand goes, one a headwind the migration narrative still hasn't priced, one a latent tailwind that mostly sits elsewhere.
For 25 years Americans flooded the Sun Belt. Cotality now shows the tide reversing: climate-driven premium spikes are eroding values and pushing net out-migration from Texas, Arizona, and Florida toward the Midwest and Northeast. First Street pegs values 17% lower in high-risk markets.
Delayed move-outs are a stored-up renter pool. Chandan and Arbor find metros with the most Gen Z still at home post steadier rents, but the deepest pools sit in secondary California and Texas markets (McAllen, El Paso), not our primary 12. A national tailwind that lands unevenly on this footprint.
The insurance line now behaves like a demand variable. Where premiums break the household budget, people leave; where the budget holds, the supply cliff does its work. Underwrite the premium curve, not just the current dollar.
It cuts both ways. Adoption is already moving the expense line and leasing velocity, and it is quietly introducing a new risk: confident answers that don't agree.
Leasing and operations AI has crossed into core infrastructure, Funnel across ~1.5M units, EliseAI inside 24 of the 25 largest owners. On a knife's-edge cycle of flat revenue and rising expenses, the operators converting demand with precision and protecting retention are the ones holding NOI.
When an owner's agent and an operator's platform define "effective rent" differently, you don't get two views of one reality, you get two numbers and a trust problem. AI's documented sycophancy compounds it; a 2026 Aalto study found AI erases the self-correction that normally checks overconfidence, every user overestimated their work, the most experienced most of all.
Everyone has access to AI now. The edge is verified, standardized context. A model will produce a fluent answer and have no idea whether to believe it; knowing whether to is still the job.
Coming into 2026, investor sentiment was broadly constructive, most expected lower rates and an improving climate. June erased the rate path the optimism was built on.
Constructive sentiment is fine; it was underwritten on cuts that aren't coming. The long-term conviction in rental housing is real, but the 2026 entry has to clear today's rate, at today's money.
The 2026 Economic Report of the President puts regulation at the center of the affordability problem, a rare alignment with the industry. The catch is buried in the fine print: a proposal that could throttle built-to-rent.
Regulation is the most durable moat in housing: every barrier that blocks new supply protects the cash flow of what is already standing. We underwrite the existing book, watch Section 901 closely for BTR exposure, and treat the supply cliff as policy-reinforced, not just cyclical.
With forward guidance gone, every release now moves the rate path on its own. Five data points between now and the July FOMC will decide whether the hawkish turn holds, and what proceeds survive an autumn refinance.
The 2-year high becomes an in-line print, and the ratio breaks below 1.0 for the first time since the 2021 recovery — the line the Fed watches to call labor "balanced" rather than tight. The hawkish hold loses its best data point, and cuts re-enter the conversation.
We underwrite a distribution of outcomes. If the jobs print cracks, cuts re-enter and pricing moves fast; if PCE stays hot, the hawkish hold extends and the refinancing wall gets taller. Build both into the model now.