What Housing Starts Are, and Why Today's Number Is 2028's Rent
The count of homes on which construction began this month, stated as a yearly pace. It is a forecast of supply, and supply is what rents and prices answer to, a year or two later.
October 28, 2026 · updated October 30, 2026
In one sentence
Housing starts are the number of new homes on which construction began in the month, counted by the Census Bureau from building permits and site visits, and stated as the annual pace that month's count would make if it continued for twelve months.
The total splits into two things that move on different clocks: single-family houses, and apartments and other buildings with two or more units. In October 2026 the total was 1.31 million at an annual rate, about 860,000 of them houses and about 440,000 apartments.
New articles and charts arrive in one short email on Monday mornings. Get the weekly digest
The analogy
Seeds in the ground. A planting report tells you about next year's harvest, not this year's dinner. A house started today is a kitchen in about seven months; an apartment building started today opens in about two years. Nothing that happens to rents or prices this month comes from this month's starts. It comes from the ground broken in 2024 and 2025.
That is why the 2022 apartment boom and the 2024–25 rent slowdown are the same event seen twice. Apartment starts hit 643,000 in late 2022, the most since 1986, as developers chased pandemic-era rents with cheap construction loans. Those buildings opened through 2024 and 2025, mostly in the Sunbelt, and rents in the cities that built the most went flat or fell. The seeds planted in 2026 are fewer: apartment starts have run around 400,000 to 500,000 this year, down a third from the peak, which is the 2028 rental market being decided now.
How it reaches your kitchen table
If you rent: in 18 to 24 months. Today's apartment starts are your lease options two years out. A city where starts are falling now will have fewer new buildings competing for tenants in 2028, whatever its rents are doing today.
If you build, or work in construction: now. Construction employment follows starts within months, and the trades, lumber yards, appliance makers and furniture stores follow the completions. Housing is the most rate-sensitive part of the economy and usually the first to turn, which is why economists watch it as an early warning.
If you're buying: through what's for sale. With most owners locked into low mortgage rates and not listing, new homes have been a far larger share of what's for sale than usual, and builders have been doing what existing sellers can't: cutting prices and buying down mortgage rates to move inventory. Single-family starts, around 900,000 a year lately, are what keeps that going.
If you own: slowly, through prices. More supply holds prices down; the lag runs years and the effect is local. The arguments below are mostly about how strong this channel is.
What moves it
Rates, first: mortgage rates for houses, construction-loan rates for apartments, both priced off the 10-year yield. Then the cost and availability of everything else: land that zoning allows building on, labor (a quarter or more of construction workers are foreign-born, so immigration enforcement reaches this number quickly), and materials, where tariffs on lumber and steel land directly. Then household formation, meaning how many people are reaching the age, income and family status where they want a place of their own. Developers decide apartment projects on five-year horizons and respond to rents; builders decide subdivisions on one-year horizons and respond to mortgage rates and the price of the last house they sold.
How to read this month's number
It is very noisy. Monthly swings of 10% are routine and weather alone can produce one; the May 2026 print was 1.18 million and June's was 1.44. Use the three-month average or compare with a year earlier, and never read one month as a turn.
Split it. Houses and apartments move on different cycles, and the total can hold steady while one rises and the other falls. The chart shows single-family starts as a line under the total; the gap between them is apartments.
Know the pipeline. Permits come before starts, completions come after, and the Census publishes both in the same release along with units under construction. Starts rising with permits behind them won't last; a large stock of units under construction means completions are coming regardless of what starts do.
Know the levels. The 2015–19 average was 1.2 million, so 2026 is running at the pre-pandemic pace. Whether that is enough depends on how big a shortage you think the country has, which is argued below; the long-run peak was 2.5 million in 1972, and the 2009 low was under 500,000.
Where the analogy breaks
A farmer who plants, harvests. A builder who starts can stop: 2008 left half-built subdivisions across Nevada and Florida, and apartment projects get shelved between permit and start when financing dries up. The harvest also arrives in different seasons by crop, seven months for a house and two years for a building, so one month's starts feed two different years. And a harvest feeds whoever buys it, while new housing mostly helps the market it's built in: Austin's new towers did nothing for Boston's rents. Starts count units, not whether anyone can afford them.
What people argue about
What the binding constraint is. Three camps, each with evidence. Zoning: the places that permit the most (Houston, Austin, Minneapolis after its 2019 reforms) built the most and saw rents cool, and the places that permit least have the highest prices regardless of rates. Rates: single-family starts fell by a third within a year of the 2022 hikes, with no change in zoning anywhere. Labor and materials: the 2025–26 drop in the immigrant construction workforce and tariffs on lumber raised costs at the same time rates were falling, and starts didn't recover. The data support all three because all three bind at different times and places, and a national number can't say which one is binding in your county.
How big the shortage is. Estimates run from about 1.5 million homes to over 4 million, depending on how household formation and vacancy are counted, and a minority of economists argue there is no national shortage at all, only a mismatch between where homes are and where jobs are. The size of the number decides whether 1.2 million starts a year is adequate or a crisis, so the estimate is the argument.
Whether building expensive apartments helps anyone else. One side points to moving chains: a new market-rate building draws tenants from older buildings, which free up units down the price ladder, and studies tracking actual moves find the chain reaches lower-income neighborhoods within a few years. The other side points to displacement and to new buildings raising the land values and rents around them. Both effects are documented. Which dominates depends on how much gets built, which brings the argument back to the first one.
See also
- Mortgage rates: the rate that moves single-family starts.
- The 10-year Treasury yield: what both kinds of construction loan are priced from.
- Inflation: where rents land in the index, a year late.
- The almanac's housing starts with every overlay.
Sources
- U.S. Census Bureau, New Residential Construction, the monthly release with permits, starts, under construction and completions.
- FRED: Housing starts (HOUST), single-family (HOUST1F), five units or more (HOUST5F), completions (COMPUTSA), units under construction (UNDCONTSA).
- Freddie Mac, "Housing Supply: A Growing Deficit" (2021), one of the shortage estimates.
- Mast, "The Effect of New Market-Rate Housing Construction on the Low-Income Housing Market" (Upjohn Institute, 2019), the moving-chain evidence.
Corrections
October 30, 2026: An earlier version defined the second group as buildings of five or more units. The apartment figures here are total starts minus single-family, which also counts buildings of two to four units.