Whose Vacancy Actually Improved

The headline says recovery

National apartment vacancy improved to 8.15% in the second quarter, down 26 basis points from the quarter before. Read on its own, that is a market finding its footing.

Read alongside the number underneath it, it is something else.

Stabilized vacancy, which strips out properties still filling up for the first time, rose 34 basis points year over year. Not in one region. In every major region of the country.

Both numbers are accurate. They are just measuring different buildings.

What is actually moving the number

The mechanism is specific enough to name. Four and five star properties absorbed more than 111,000 units in the second quarter, roughly 68% of all national absorption. New supply, filling up.

When a new building leases up, its vacancy falls fast, and the national average falls with it. But those renters came from somewhere. In most cases they came out of existing communities, which is why the stabilized number moved the opposite direction at the same time.

Austin is the sharpest illustration. Vacancy across its four and five star properties fell 398 basis points year over year. Its stabilized vacancy rose. One market, two stories, and only one of them makes the headline.

If you own a stabilized apartment community, the national recovery number is not describing your building.

Fifteen of the fifty largest apartment markets showed the same split: overall vacancy down, stabilized vacancy up.

Source: CoStar second quarter 2026 data, as reported by CRE Daily, August 13, 2026.

Running the same test on El Paso

Any test worth applying to someone else's market is worth applying to your own. So here is the honest version.

The mechanism above needs one specific ingredient to work: new, higher end product coming online and pulling renters out of existing buildings. Take that ingredient away and the distortion has nothing to run on.

El Paso holds 8,233 Class A units. Zero were delivered over the trailing twelve months. Zero are under construction today. Across the entire metro, 570 units are under construction against a standing inventory of 47,611, which works out to 1.2%.

Source: CoStar Multi-Family Market Report, El Paso TX, July 2026. Figures approximate.

El Paso vacancy fell a full point to 5.3% over the same stretch, with 992 units absorbed against 628 delivered.

One methodological note, because the distinction is the whole point of this piece. CoStar publishes an overall vacancy figure for El Paso, not a separate stabilized one, which makes this an inference rather than a direct reading. It is a tightly constrained one. The gap between overall and stabilized vacancy can only be as wide as the lease up inventory allows, and El Paso's lease up inventory is 628 units delivered across twelve months against a 47,611 unit base, none of it Class A. There is not enough new product in that market to open a divergence anywhere near the size of the national one.

So the comparison is worth making directly. Austin has enough new supply filling up to move its headline vacancy 398 basis points while its stabilized owners lose ground. El Paso has almost no such inventory to move. On the dimension this article measures, a stabilized owner in El Paso is competing against nearly nothing that has not already been built. A stabilized owner in Austin is competing against a great deal of it.

None of which makes the market easy

Supply is one variable. It tells me where the pressure on operations is coming from, and in El Paso it currently points in a favorable direction.

Debt answers a different question. It tells me how much room there is to be wrong about the first one. And it moved this month in a direction worth flagging.

Multifamily lending has gone from scarce to crowded. Bank lending is up 30% year over year. Banks are beating the agencies by 30 to 40 basis points on some deals. For a disciplined buyer, more competition among lenders is a good thing.

Here is the part that stopped me. Cushman & Wakefield reports that some debt funds which once charged 1% to 3% of a loan balance for an extension may now ask closer to 10%. The context is specific: borrowers coming back for another modification, largely developers carrying maturing construction loans while they finish stabilizing an asset.

That is not my segment. Nobody is extending a construction loan on a 1985 vintage workforce property. But the direction of the number carries over regardless of who is paying it. Ten percent of a loan balance is not a fee. It is the price of having been wrong about timing, and that price is moving.

The rate tailwind has largely stopped as well. All in borrowing costs fell just 4 basis points in the second quarter. The comparable drop in the fourth quarter of 2025 was 45.

Capital markets remain the part of this business where I am a student rather than an authority. But that number changes how a deal reads. Debt service coverage ratio is the first metric I run, ahead of rent comps and ahead of the value add plan. My floor is 1.40x on in place income, not on projected income. A deal that only clears 1.40x after the business plan works is a deal that depends on the business plan working.

That floor is not there to make a model look conservative. It is there because needing more time keeps getting more expensive.

Sources: Cushman & Wakefield and CBRE data, as reported by CRE Daily, August 25, 2026. Altus Group data, as reported by CRE Daily, August 24, 2026.

One question worth sitting with

Every market has a headline number that gets quoted and a second number that explains it. The work is knowing which is which before the capital goes in.

When you are evaluating a market, what number do you check first, and what have you learned it hides?