America's Apartment Overhang Map

• 

  Newly built mid-rise apartment building, the Class A supply that drove 2024 deliveries

The country delivered more apartments in 2024 than in any year since 1986, and then stopped building. The vacancy that followed is being read as a supply failure. Half of it is a demand failure, and it is concentrated in about a dozen metros.

In June 2025, the vacancy rate on two-bedroom apartments in Austin reached roughly 10 percent. Four years earlier it had been 3.96 percent.

Rents in the metro were down about 17 percent from their August 2022 peak. Austin finished 2025 down 4.3 percent year over year, the worst performance of any major American market, capping eight consecutive quarters at the bottom of the national table. In some submarkets landlords were offering two months free.

That is the image the phrase "apartment oversupply" now conjures, and Austin earned it.

Then look at what happened next. In the first quarter of 2026, Austin absorbed more than 20,000 units against roughly 14,900 delivered. Demand exceeded supply. The metro that everyone points to as the definitive overbuilding disaster had already flipped.

The correction is running faster than the narrative, which is the first of several things about the American apartment market in 2026 that the headline numbers get wrong.

Which number is the vacancy rate?

Before any of this can be discussed sensibly, one problem has to be dealt with, because it invalidates a great deal of published commentary.

There is no single national apartment vacancy rate. There are at least five, and they disagree by more than four hundred basis points.

These are all reputable, carefully constructed series, and they are not contradicting each other. They are counting different buildings. The Census survey covers every rental unit in the country, single-family rentals, duplexes, small walk-ups, the accessory apartment over somebody's garage. CBRE and RealPage count stabilized institutional apartment stock, which is a much narrower and much better-performing universe.

The Census data is also more useful than it usually gets credit for, because it disaggregates. In the fourth quarter of 2025, rental vacancy ran 9.1 percent across the South and 5.5 percent across the West. Principal cities showed 8.0 percent, suburbs 6.9 percent, and areas outside metropolitan statistical areas 5.8 percent.

Hold that South figure. It matters later.

What actually got built

Analytics.loan's analysis of Census construction data shows 608,000 multifamily units completed in 2024, the most since 1986. Completions fell to 484,000 in 2025. The South alone delivered 217,000 units in 2025, down from 292,000 the year before.

The starts data is where the story turns. Roughly 55,000 multifamily units broke ground in the first quarter of 2025, the lowest quarterly volume since 2011. Annualized starts bottomed near 316,000 in May before recovering to 402,000 by December. Units under construction fell from more than a million in December 2023 to between 670,000 and 690,000 two years later, and by one count the number underway is down roughly 50 percent from the 2023 peak. The National Association of Home Builders projects 392,000 starts in 2026 and 367,000 in 2027.

The lag from groundbreaking to certificate of occupancy runs 19 to 22 months. Which means 2027 and 2028 deliveries are already largely determined, and they are determined by the weakest starts year since 2011.

The pipeline that created the overhang has been switched off. The overhang itself has not cleared yet, because concrete poured in 2023 is still being finished.

The map

The second thing the national numbers obscure is that this is a regional event with a very clear geographic logic.

Metro	Supply signal	Rent growth, YoY
Austin	2BR vacancy ~10%, from 3.96% in Sept 2021; rents −17% from peak	−4.3%
San Antonio	9.26% vacancy	~−5%
Denver	Heavy 2023-24 delivery cohort	−3.3%
Phoenix	Completions −43% in 2025; starts −52% in 2024	−2.6%
Dallas-Fort Worth	8.66% vacancy	Negative
Houston	8.46% vacancy	Negative
Nashville	13,300 units underway = 7.3% of inventory	Negative
Charlotte	22,000 underway = 9.6% of inventory; record 16,700 delivered 2024	Negative
Miami	12.6% of inventory underway, highest nationally	Mixed
Chicago	Adding 1-2% to stock	+3.8% to +4.6%
Cincinnati	Thin pipeline	+2.8%
Norfolk	Thin pipeline	+2.8%
New York	Adding 1-2% to stock	+2.5%
Pittsburgh	Thin pipeline	+2.5%
Kansas City	Thin pipeline	+2.4%
Cleveland	Thin pipeline	+2.3%
Detroit	Thin pipeline	+2.3%

Chicago led every major American metro on rent growth in 2025. The Midwest as a region ran 2.0 percent year over year, the strongest in the country. PwC and the Urban Land Institute describe New York and Chicago as seriously undersupplied, each adding only one to two percent to standing stock.

The two lists sort almost perfectly by how hard it is to get a building permit. Metros with permissive entitlement, cheap land and by-right multifamily zoning built into the teeth of a demand forecast. Metros where approval takes three years and land costs four times as much did not, and are being rewarded with rent growth in a year the industry has spent calling an oversupply year.

One honest complication. The Census rental vacancy rate for the entire South region is 9.1 percent, not just for a handful of headline metros. That is broader than a dozen-metro story. The concentration thesis is directionally correct and geographically fuzzier than a clean metro list suggests, and anyone underwriting a secondary Southern market should not assume the softness stops at the boundaries of the metros that make the news.

The record absorption problem

Here is the fact that most complicates the oversupply narrative, and it is rarely reported alongside the vacancy numbers.

2024 was a record absorption year. RealPage counted 666,000 units absorbed; CBRE counted 530,600. Either figure is more than double 2023. Absorption in 2025 peaked near 785,000 units annualized in the second quarter before moderating.

Renters were signing leases at a pace the industry had never seen. Vacancy rose anyway.

This is important because it changes the diagnosis. A market with collapsing demand and rising vacancy has a structural problem. A market with record demand and rising vacancy has a timing problem. The apartment sector in 2024 and most of 2025 was the second kind.

And then it became something else.

The denominator moved

The demand line did not hold. It fell, hard and fast, for a reason that has almost nothing to do with real estate.

Net international migration to the United States ran 2.7 million in 2024. It fell to 1.3 million in 2025, and the Census Bureau's Vintage 2025 estimates, released in January 2026, project approximately 321,000 for 2026 if current trends continue. The Bureau's assistant division chief for population estimates, Christine Hartley, described the drop from 2.7 million to 1.3 million and the projected further decline in those terms.

Independent estimates are more dramatic still. Brookings puts 2025 net migration somewhere between negative 295,000 and negative 10,000, the first negative reading in more than fifty years, and projects a 2026 range from negative 925,000 to positive 185,000. John Burns Research and Consulting's Chris Finnigan put the 2025 decline at roughly 82 percent, the lowest level in more than forty years.

The household consequence is direct. Total US household growth slowed to about 1.1 million in 2025, against roughly 2 million a year during the pandemic surge. Renter formation followed the same curve with a lag: a record 784,000 apartment households added in the year to the second quarter of 2025, decelerating to 366,000 by the fourth quarter. Harvard's Joint Center for Housing Studies estimates that switching from baseline to low-immigration projections removes 74,000 to 86,000 renter households a year through 2035.

Now run the arithmetic that the vacancy charts imply but never state. Completions of 484,000 in 2025. Renter formation exiting the year at an annualized 366,000. A gap of roughly 118,000 units a year, and that gap was not opened by developers.

Five states accounted for 47.9 percent of 2025 net international migration: Florida, Texas, California, New York and New Jersey. Three of those five contain most of the metros at the top of the overhang table.

This is a materially different diagnosis than the one in circulation. The standard account is that developers misjudged demand. The data says the demand assumption they underwrote against was accurate when they underwrote it, and was revised down by well over a million people a year after the concrete had been poured. A building started in mid-2023 was capitalized against a 2.7-million-person immigration run rate and delivered into a 321,000-person one.

No underwriting model catches that. It is not a real estate error.

The class story, corrected

The popular version of the class narrative holds that shiny new luxury product is the distressed segment while workforce housing holds up. The data says that is right on levels and wrong on direction.

But Class A absorbed roughly 75 percent of every unit leased in that quarter, about 115,000 of 151,440. Stabilized Class A occupancy reached 95.7 percent in May 2025, a two-year high, and average days-vacant fell to 25 from 32 in December 2023. The lease-up is working.

It is working expensively. Class A concessions in several Sun Belt metros are at their highest levels since the financial crisis, with six to eight weeks free now routine. Charlotte shows the endpoint of that dynamic cleanly: Class A occupancy at 85.5 percent against Class B at 92.3 percent.

And the concessions are how the damage transmits downward. When a brand-new building offers two months free, its effective rent lands in the same range as stabilized Class B asking rents. The concession is a Class A expense and a Class B pricing problem. The segment with the better occupancy figures is the one losing the ability to raise rent.

Where the distress actually is

Multifamily CMBS delinquency reached 7.12 percent in October 2025, up 53 basis points in a single month, breaking 7 percent for the first time since December 2015 and running nearly double the 3.24 percent recorded six months earlier. KBRA identified multifamily as the highest-volume newly distressed sector that month, with $669 million representing 39.4 percent of all new distress.

Set that against the agency books. Fannie Mae's 60-plus-day multifamily delinquency rate was 0.78 percent in the first quarter of 2026. Freddie Mac's was 0.43 percent. FHA-insured multifamily stood at 0.33 percent in December 2025.

Those are not small differences. They are nearly an order of magnitude, on broadly similar collateral.

The reconciliation is vintage, not asset quality. The distress sits almost entirely in 2021 and 2022 floating-rate bridge debt. Tides Equities, with roughly $7 billion in assets and about 15,000 Texas units, has lost properties to foreclosure across Dallas-Fort Worth and Las Vegas, with its principals personally sued. GVA had roughly 80 percent of its portfolio distressed by late 2023 and more than $413 million in defaults. Lurin Capital has defaulted on at least $710 million and entered receivership amid code violations. Ashcroft Capital called nearly 20 percent additional capital from investors.

The mechanism was never complicated. Bridge debt priced off SOFR when SOFR was 0.05 percent, on value-add business plans that carried no stress test for a 500-basis-point move. The rate caps expired. Nothing failed in the buildings.

Roughly $162.1 billion of multifamily loans mature in 2026 and $167.7 billion in 2027.

The lender that did not step back

While private bridge capital was blowing up and banks were retreating from construction, the federal channels expanded, and that is where the supply of 2028 is being financed right now.

The Federal Housing Finance Agency set 2026 multifamily purchase caps at $88 billion each for Fannie Mae and Freddie Mac, $176 billion combined, up 20.5 percent from $73 billion each in 2025. At least half of that must be mission-driven, and workforce housing sits outside the caps entirely. That is a deliberate counter-cyclical expansion into a market the private sector had decided to avoid, and it is the largest single reason the agency delinquency numbers look nothing like the CMBS numbers: the agencies never left, so they never had to be replaced by anyone worse.

HUD's insured multifamily programs tell a more specific story. In fiscal 2025, FHA endorsed 434 multifamily loans totaling $9.8 billion. Of that, the Section 221(d)(4) construction and substantial rehabilitation program accounted for 117 loans and $3.36 billion, with Section 223(f) refinancing taking 221 loans and $4.27 billion.

That construction figure is the one worth watching. In a year when private construction lending largely stopped, the federal insurance program wrote $3.36 billion of new multifamily construction debt, forty-year, fixed-rate, non-recourse, at leverage no bank would offer. The demand for it exceeded what got closed: firm commitments totaled $14.1 billion against $20.9 billion in applications.

The constraint is administrative rather than financial. HUD's multifamily staffing has been cut roughly 30 percent from 2023 levels, and during the 2025 federal shutdown the agency continued closing deals that already held firm commitments but stopped accepting new applications altogether. The department streamlined its multifamily environmental review process in May 2025, which helps, but a program whose bottleneck is underwriter headcount cannot scale to meet a supply cliff on demand.

The rural channel deserves a mention it almost never gets. USDA's Section 538 guaranteed rural rental housing program ran roughly $230 million a year across 96 to 150 loans in fiscal 2020 and 2021, rising toward a $400 million program level by fiscal 2023 and 2024. Cumulatively it has supported about 51,000 units and guaranteed nearly $1.5 billion since inception.

Its delinquency rate is under 0.5 percent.

Set that beside 7.12 percent on multifamily CMBS and the comparison is instructive. The rural program lends at lower leverage, to sponsors who intend to hold, on fixed-rate terms, in markets nobody was speculating in. It produced almost no losses through the worst multifamily credit cycle since the financial crisis. The difference between those two numbers is not geography or asset quality. It is capital structure and sponsor intent, which is the same lesson the bridge-debt vintage taught expensively.

The cost line nobody underwrote

One operating expense deserves separate treatment because it has done more damage to net operating income than vacancy has.

The Federal Reserve Bank of Minneapolis found that more than half of all multifamily operating-expense inflation since 2020 is attributable to property insurance alone. Premiums roughly doubled between 2021 and 2024, with successive annual increases of about 14 percent, then 22 percent, then 45 percent. Yardi put insurance at approximately $636 per unit in early 2025, up 27.7 percent year over year and 129 percent since 2018.

That cost arrives regardless of what rents do, and it explains why NOI has fallen at operators whose occupancy looks fine. MAA's same-store net operating income declined 1.4 percent in 2025 and Equity Residential's 0.9 percent, while AvalonBay's grew 1.9 percent. The split is Sun Belt against coastal, which is the same map again.

The lease data shows the squeeze at the unit level. MAA reported new-lease rents down 5.2 percent in the third quarter against renewals up 4.5 percent. Existing residents are absorbing increases. New residents are refusing to pay asking rent. AvalonBay's chief executive, Ben Schall, told investors in February that the company's 41 percent turnover rate in 2025 was the lowest in its history, which is a way of saying nobody could afford to move.

Does anything pencil?

The question every developer is being asked in 2026 is whether new multifamily construction works anywhere, and the answer is more interesting than a flat no.

AvalonBay started $1.65 billion of development in 2025 at a projected 6.2 percent yield on cost. That is a real number from a real balance sheet, and it is not a distressed one, it is the kind of spread that justified development throughout the last cycle.

Note where AvalonBay builds. It is a coastal and inner-suburban operator whose markets are almost perfectly the inverse of the overhang table. Development pencils in the places that did not overbuild, because rents there are still growing, land is expensive enough to deter marginal competitors, and entitlement difficulty functions as a moat. It does not pencil in Austin or Charlotte, where a developer would be delivering into standing concessions.

That is the same map a third time, viewed from the capital side. The metros with the worst current fundamentals are the ones where nobody will start anything, which is precisely why they will be short first.

There is a second constraint on any supply response, and it is one almost nobody in real estate priced.

The immigration reversal did not only remove renters. It removed the people who build apartments. Foreign-born workers are substantially over-represented in construction trades, HUD's own analysis and Congressional Research Service work both note the concentration. The same policy shift that cut net international migration from 2.7 million to a projected 321,000 also thinned the framing, drywall, concrete and finish crews on which any delivery forecast depends.

That cuts in an unexpected direction. It makes the demand shortfall worse in the near term and the supply shortfall worse in the medium term. A market that loses renters and builders simultaneously does not simply re-equilibrate at a lower level; it overshoots twice, in opposite directions, several years apart.

Layer insurance on top. Property insurance now consumes a share of operating expense that did not exist in any pre-2021 pro forma, and for a development underwriting exercise it lands in the stabilized year rather than the construction budget, which means it comes straight out of yield on cost. A project penciling at 6.2 percent today would have penciled nearer 7 percent on 2019 insurance assumptions, holding everything else equal.

None of that shows up in a vacancy chart. All of it shows up in what gets started, which is the number that determines 2028.

The cliff

And the delivery collapse is steepest in precisely the metros that overbuilt. Austin completions are down about 60 percent, Phoenix 43 percent, Nashville 41 percent, and Charlotte starts 40 percent. The markets currently offering two months free will be the first ones short.

The industry's long-run demand estimate, 4.3 million additional apartments needed by 2035, roughly 266,000 a year plus a standing 600,000-unit deficit, should be handled with care. It was commissioned in 2022 against an immigration assumption that has since been revoked, and its author has not published a revision reflecting the collapse.

But the direction survives the caveat. A country building 367,000 units a year against even a reduced formation rate, in metros where entitlement is easy and land is cheap, will run short before it runs long.

What we are watching

Five measurable things will settle whether this reading holds.

Net international migration in the next Census vintage is the first and most consequential. If it prints materially above the projected 321,000, the demand-shortfall half of this thesis weakens considerably.

Multifamily starts are the second. Sustained readings below roughly 350,000 annualized through 2026 make the 2028 shortage close to arithmetic.

The vacancy peak is third. If provider vacancy fails to top out by early 2027 as forecast, the "cyclical, not structural" framing is wrong and the demand impairment is deeper than the migration data alone explains.

Sentiment is fourth, and it is currently pointing the wrong way. The National Multifamily Housing Council's January 2026 survey of 98 executives produced a Market Tightness Index of 32, below the 50 breakeven, and looser for a second consecutive quarter, even as headline vacancy ticked down for the first sustained stretch in four years. The people running these portfolios are not yet convinced.

The fifth is whether distress migrates. So far it has stayed inside the 2021 and 2022 bridge vintage. If it appears in stabilized, agency-financed, well-located assets, the problem is no longer a capital-structure story.

Until those move, the accurate description of the American apartment market is not that the country overbuilt. It built at a high but historically defensible rate against a population forecast that was revised down by more than two million people after construction began, in a set of metros where building is easy, while the metros where building is hard posted the strongest rent growth in the country.

That is a different problem with a different remedy. It resolves on a construction timetable, not a policy one, and the timetable says 2028.

Methodology

Completions, starts, units under construction and rental vacancy figures are Analytics.loan's analysis of Census Bureau data, including the Survey of Construction, New Residential Construction, and the Housing Vacancy Survey, through the most recent 2026 releases. Readers should note that the second-quarter 2026 Housing Vacancy Survey was affected by a lapse in federal funding for data collection.

Vacancy rates are reported as a range rather than a single figure because the major series measure different housing universes: the Census Housing Vacancy Survey covers all rental housing (7.3 percent, Q2 2026) and CBRE and RealPage cover stabilized institutional apartment stock (approximately 4.6 percent, Q3 2025). These are not reconcilable and should not be blended.

Net international migration and household formation figures are from the Census Bureau's Vintage 2025 population estimates, released January 2026, with independent ranges from Brookings and the Harvard Joint Center for Housing Studies. CMBS delinquency figures are from Trepp and KBRA; agency delinquency from Fannie Mae, Freddie Mac and FHA disclosures. Operating expense and insurance figures are from the Federal Reserve Bank of Minneapolis and Yardi Matrix. Operator results are from AvalonBay, Equity Residential and MAA public disclosures. Forward figures, 2026 and 2027 deliveries, starts forecasts, vacancy projections and the 2035 demand estimate, are forecasts, not results, and the 2035 estimate carries a 2022 base year predating the immigration reversal.

Key figures for citation

Metric	Value
Multifamily completions, 2024	608,000 units, most since 1986
Multifamily completions, 2025	484,000 units
Q1 2025 multifamily groundbreakings	~55,000 units, lowest since 2011
Units under construction, Dec 2025 vs Dec 2023	670,000-690,000 vs over 1,000,000
National rental vacancy, range across sources	4.6% to 8.8%
Census rental vacancy, Q2 2026	7.3%
Census rental vacancy, South region, Q4 2025	9.1%
Net international migration, 2024 / 2025 / 2026 projected	2.7m / 1.3m / ~321,000
US household growth, 2025	~1.1m (vs ~2m/yr pandemic pace)
Renter household formation, Q2 2025 vs Q4 2025	784,000 → 366,000 annualized
Units absorbed, 2024	666,000 (RealPage) / 530,600 (CBRE)
Q2 2025 supply vs absorption	175,655 added vs 151,440 absorbed
Class A vs Class B vacancy, Q2 2025	11.5% vs 7.5%
Class A share of quarterly absorption	~75%
Best major-metro rent growth, 2025	Chicago, +3.8% to +4.6%
Worst major-metro rent growth, 2025	Austin, −4.3%
Multifamily CMBS delinquency, Oct 2025	7.12%, first above 7% since Dec 2015
Fannie / Freddie / FHA multifamily delinquency	0.78% / 0.43% / 0.33%
Share of multifamily opex inflation from insurance	Over 50%
Multifamily loan maturities, 2026 / 2027	$162.1bn / $167.7bn

Frequently asked questions

What is the US apartment vacancy rate?

There is no single figure, and the major series disagree by more than 400 basis points because they count different buildings. The Census Housing Vacancy Survey put national rental vacancy at 7.3 percent in the second quarter of 2026 across all rental housing. CBRE and RealPage, measuring stabilized institutional stock, showed roughly 4.6 percent.

Is there a national apartment oversupply?

It is concentrated rather than national. Austin, San Antonio, Denver, Phoenix, Dallas-Fort Worth, Houston, Nashville, Charlotte and Miami carry the pipeline and the falling rents. Chicago led every major metro on rent growth in 2025 at 3.8 to 4.6 percent, with Cincinnati, Norfolk, New York, Pittsburgh, Kansas City, Cleveland and Detroit all positive. The two lists sort almost perfectly by how hard it is to get a building permit.

How much multifamily was built in 2024 and 2025?

Census data shows 608,000 multifamily units completed in 2024, the most since 1986, falling to 484,000 in 2025. Roughly 55,000 units broke ground in the first quarter of 2025, the lowest quarterly volume since 2011, and units under construction fell from more than a million in December 2023 to between 670,000 and 690,000 two years later.

How has immigration affected apartment demand?

Substantially. Net international migration ran 2.7 million in 2024, fell to 1.3 million in 2025, and is projected at approximately 321,000 for 2026. Total household growth slowed to about 1.1 million in 2025 against roughly 2 million a year during the pandemic surge, and renter formation decelerated from a record 784,000 annualized in mid-2025 to 366,000 by the fourth quarter. A building started in 2023 was capitalized against a 2.7 million run rate and delivered into a 321,000 one.

When will apartment vacancy peak?

Current forecasts put the peak in early 2027 near 8.8 percent, easing to 8.4 percent by year-end. The start-to-completion lag runs 19 to 22 months, so 2027 and 2028 deliveries are already fixed by the weakest starts year since 2011. The delivery collapse is steepest in the metros that overbuilt, which means they are positioned to be short first.

Why are apartment operating costs rising?

Insurance, primarily. Federal Reserve research found that more than half of all multifamily operating-expense inflation since 2020 is attributable to property insurance alone, with premiums roughly doubling between 2021 and 2024 through successive annual increases of about 14 percent, 22 percent and 45 percent. Insurance reached approximately $636 per unit in early 2025, up 129 percent since 2018.

Analytics.loan produces lender-grade feasibility and market analysis for SBA, USDA and conventional commercial real estate credit.

Sources:

• U.S. Census Bureau, Survey of Construction, New Residential Construction and Housing Vacancy Survey, through 2026 releases

• U.S. Census Bureau, Vintage 2025 population estimates, net international migration and household formation

• Brookings Institution and Harvard Joint Center for Housing Studies, independent migration and renter formation estimates

• CBRE, absorption and stabilized apartment vacancy

• Trepp and KBRA, multifamily CMBS delinquency and distress

• Fannie Mae, Freddie Mac and FHA, agency multifamily delinquency disclosures

• Federal Housing Finance Agency, 2026 multifamily purchase caps

• U.S. Department of Housing and Urban Development, FHA insured multifamily endorsements

• USDA Rural Development, Section 538 guaranteed rural rental housing

• Federal Reserve Bank of Minneapolis, multifamily operating expense and insurance analysis

• AvalonBay, Equity Residential and MAA public financial disclosures

• National Association of Realtors, existing-home sales