Apartment Completions Just Fell 35.7%. The Rent Glut Everyone Is Still Underwriting Is Already Over.

The builders' headline this morning was a number that might be zero. The real one was buried three tables down: new apartments hitting the market fell 35.7% in a year.

At 8:30 this morning the Census Bureau and HUD published August's new residential construction report. The line most people will repeat is single-family starts up 7.6%. The same sentence carries a confidence interval of plus or minus 14.0%, which by the Bureau's own rule means it can't say whether starts rose at all. The number underneath it says the rent glut investors are still underwriting is already over, and I'll put a date on that below.

The number that cleared the bar sits further down. Buildings with five or more units finished at an annual rate of 302,000 in August, down from 470,000 a year earlier.

That's the lowest reading in the report's thirteen-month table, and at plus or minus 19.6% the drop is statistically real.

This lands one day after the Federal Reserve raised rates a quarter point to 3.75% to 4%, by a 12-0 vote. The standard rental pro forma still assumes the apartment glut of 2024 and 2025 keeps rents flat for years. I think that underwriting is a year late, and the supply numbers are why.

The supply wave broke while everyone watched starts

Rents follow deliveries, not starts. A tower that breaks ground today competes for tenants in 2028. A tower that finishes today competes this winter. So if you want to know where rents go over the next twelve months, completions are the number to watch, and completions are falling.

Here is the pipeline from the Census tables:

The last row is the tell.

Developers are holding 128,000 permitted apartment units they haven't broken ground on, up 15.3% in a year. A permit is an option, not a building. When debt is cheap, people exercise the option. When the Fed is raising rates, they wait.

The construction lenders know it. After Wednesday's decision, FMI chief economist Brian Strawberry told Construction Dive that a hike "would immediately raise the cost of floating-rate construction loans." He also drew the line from there to rents:

"That squeeze plus costly construction debt is thinning the pipeline of new starts. Fewer deliveries eventually mean tighter vacancies and the next round of rent growth and building."

Tenants noticed before investors did

If the supply story were only theory, I'd hold it loosely.

It isn't. The tenant side of the market has started to turn.

Apartment List's August national rent report puts the median rent at $1,390. That's still down 0.8% from a year ago, which is the number the rents-are-soft crowd quotes. What they skip is the direction. August was the seventh straight month of increases, and the national vacancy index slipped to 7.1% from a February peak of 7.3%. Apartment List calls it the first decline in that index since late 2021.

So a market still absorbing the 2024 construction peak, which Apartment List puts at more than 600,000 new units, has started filling units while new deliveries fall by a third. You don't need a forecasting model for that.

You need arithmetic.

The regional split matters for where you buy. Completions across all building types fell 21.8% year over year in the South and 31.3% in the West, both outside the margin of error. They are also where the weakest rent numbers sit today: San Antonio is down 5.1% year over year and Austin is down 2.9%, the two steepest declines in the Apartment List report.

That's the setup I like. The markets with the worst trailing rent numbers are the ones losing new supply fastest.

The 666,000 units still in the pipe

So isn't the glut just delayed? Partly, and I don't want to oversell this. The Census report shows 666,000 apartment units still under construction in August, only 2.5% fewer than a year ago, and that change isn't statistically significant. Units still in progress will keep delivering into 2027.

But under construction isn't the same thing as arriving next month. When a completions rate falls this hard while the under-construction count barely moves, projects are probably taking longer to finish. That would fit builders slowing work on marginal deals as financing gets more expensive. A slower drip of deliveries helps landlords whose buildings are already leasing.

The other honest caveat: August is one month, and the Bureau's own notes say it can take six months to confirm a trend in completions. The year-to-date figure, down 15.7%, is the steadier read, and it points the same way.

Common Mistakes Investors Make Here

• Reading starts as supply. Starts tell you about 2028. Completions tell you about this winter. The single-family starts headline tells you almost nothing, because the Bureau can't distinguish it from zero.
• Underwriting trailing rent growth forward. A San Antonio pro forma built on minus 5.1% rent growth prices in a glut that is already shrinking. The seller's broker won't correct you.
• Waiting for rates to fall before buying. The same rate hike that makes your mortgage more expensive is keeping 128,000 permitted apartments from getting built. Higher rates are part of why the supply picture looks better.
• Treating national vacancy as your vacancy. A 7.1% national index averages San Francisco, up 26% in rent, with Austin. Pull the submarket.

Where I Land

I think the apartment glut is already over as a pricing story, and most rental investors will realize it a year after the data did. If I were buying rentals with my own money this fall, I'd be shopping the Sun Belt metros with the ugliest trailing rent numbers, San Antonio and Austin first, and underwriting flat rent for twelve months and normal growth after that, not the negative trailing numbers that make those buildings look expensive today. Here's the dated call: Apartment List's national year-over-year rent figure will be positive by its August 2027 report. If it isn't, I'll write that post too.

How to Use PropGPT for This

"Show me 2-4 unit and small multifamily properties in San Antonio, TX listed for 60+ days or with at least one price reduction, and include estimated rents for each."

This finds sellers who priced into last year's rent weakness and haven't found a buyer, which is where negotiating room lives.

"For [address], compare the current asking rent in its ZIP code to comparable rentals and estimate cash flow at a 7% mortgage rate with 25% down, assuming 0% rent growth in year one and 3% in year two."

This makes you underwrite the thesis on purpose: flat now, recovery later, at today's rate instead of the one you're hoping for.

"List absentee-owned 5-20 unit apartment buildings in Austin, TX that the current owner has held for more than 10 years."

Long-hold owners who sat through the supply wave are often tired of it. A clean offer on a flat-rent underwrite is a real conversation with them.

"Pull the last 12 months of sales for small multifamily properties within one mile of [address] and show price per unit and days on market."

This checks whether the local market has already priced in the turn. If price per unit is still falling while listings sit, you're early, which is where you want to be.

The Bottom Line

This morning's report gave the market a noisy single-family headline and a clean apartment number. The clean number says new apartments finished at a 35.7% slower pace than a year ago, developers are sitting on 15.3% more unbuilt permits, and the Fed just made the next start more expensive.

Buy where the trailing rent data looks worst and the pipeline is emptying fastest, and underwrite flat before you underwrite growth.

The Bureau says it can take six months to confirm a completions trend, and I'm calling this one after one. If Apartment List prints a negative number next August, I'll be the investor who bought the recovery too early.