New Home Sales Rose 6.4% in August to a 684,000 Annual Rate, the Highest Since December 2025 and Above Forecasts
The Census Bureau's estimate was 2.0% below August 2025. The monthly gain fell within the survey's margin of error. Builders cut prices while 30-year mortgage rates averaged 6.95%.
A 6.4% Jump With a 19.5% Margin of Error
Americans bought new single-family homes at a seasonally adjusted annual rate of 684,000 in August 2026. That's according to figures the Census Bureau and the Department of Housing and Urban Development released on September 24[1][2]. It was 6.4% above July's rate and the fastest pace since December 2025[1][2]. It also beat economists' forecasts, which had clustered around 615,000 to 620,000[2][8][9].
Here's the catch. The Census Bureau builds its new-home sales number from a sample of builder contracts, not a full count. That means every monthly change comes with a margin of error, and this month's is ±19.5%[3]. A 6.4% gain sitting inside a 19.5% error band isn't a confirmed trend. It's a number that could be a modest rise, a modest fall, or something bigger than reported, and the agency says so itself.
The July figure moved too. It was first reported at 607,000 and has since been revised up to 643,000[3][8]. That revision alone accounts for a good chunk of why August looked like such a strong beat against forecasts built on the old, lower July number. None of this makes the 684,000 figure wrong. It just means the "jump" headlines ran ahead of what the data can actually support.
Same Report, Three Different Stories
Strip away the spin and a few things are simply true. New-home sales in August were about 2.0% below where they stood a year earlier, in August 2025[1]. The median price of a new home was $393,700, up slightly from July but down 5.8% from a year ago[1][6]. Mortgage rates averaged 6.95% on a 30-year fixed loan, the highest since January 2025[2][7]. And builders were sitting on 483,000 homes for sale, with 113,000 of those already finished and empty[6].
From there, the story splits three ways, and each version leans on facts nobody disputes. Reuters and Bloomberg both framed it as a genuine beat: sales rose as builders cut prices, which reads as demand responding to lower costs[2][4]. The trade outlet HousingWire told a more skeptical version, pointing to reports that 80% to 90% of new-home sales now require a mortgage-rate buydown, where the builder pays money upfront to lower the buyer's interest rate[5]. Lennar, one of the largest builders, said roughly half of the prospective buyers touring its communities couldn't qualify for a mortgage at all[5]. And the homebuilders' own trade group, the National Association of Home Builders, framed the same data as proof that high rates are strangling a market that would otherwise be thriving[7].
What a "Buydown" Actually Buys
The word doing the most work in this dispute is "buydown," and it's worth slowing down on. When a builder offers a buydown, it pays the mortgage lender a lump sum upfront so the buyer's monthly interest rate comes in lower than the market rate, either for the life of the loan or just the first few years. The buyer sees a smaller monthly payment. The builder, in effect, is cutting the price of the home, just in a form that doesn't show up in the sale price or the median-price statistic.
That distinction is why the same 684,000 figure can be read as either strength or subsidy. If most sales are happening because builders are eating the cost of a buydown, the headline number measures how much builders are willing to spend to move inventory, not how many buyers could afford a house on their own terms[5]. If sales are happening because builders cut sticker prices and buyers are responding, that looks more like ordinary market adjustment[2].
Builders have a clear reason to prefer buydowns to price cuts. Cutting the listed price too aggressively can drag down prices on unsold inventory nearby and on future phases of the same development. A buydown lets a builder move a home without permanently marking down what the next one is worth. But the ordinary lever is running out of margin, according to HousingWire's reporting, which noted that Cotality economist Selma Hepp put the buydown share of sales at 80% to 90%[5].
Why Neither Side Controls the Real Problem
Underneath the framing fight sits a structural mismatch that neither builders nor buyers can fix on their own. Builders are carrying 483,000 unsold homes, with 113,000 of them already finished and costing money every month in loan interest, taxes, and upkeep[6]. That gives them a straightforward incentive to sell now, whatever it takes. Buyers, meanwhile, are facing a 30-year mortgage rate of 6.95%, the highest since January 2025[2][7].
Reuters reported that rates have climbed nearly a full percentage point since the U.S.-Israeli war with Iran began at the end of February 2026, as the conflict pushed up energy prices and long-term Treasury yields[2]. Neither builders nor buyers set that rate. It's a global market response to a war, not a decision either side in this dispute can undo. Builders can offer buydowns to soften the effect on any one buyer, but they can't lower the underlying rate market-wide, and doing so for every buyer would eat further into their margins.
At 8.5 months of supply, the current inventory would take most of a year to sell off at August's pace[1][6]. The common rule of thumb puts a balanced market at around six months of supply. That gap is why builders have both the incentive and, for now, the room to keep offering incentives rather than let homes sit unsold.
The Headlines Didn't Agree Either
Coverage of the same report split along fairly predictable lines. Reuters led with "jump" and "surged," though the body of its story credited price cuts and incentives and flagged rising mortgage rates as a drag — it just didn't mention the margin of error[2]. Bloomberg's headline, "US New-Home Sales Rise to Fastest This Year as Prices Drop," paired the gain with falling prices but left out the year-over-year decline[4].
HousingWire went the other direction, framing the release around builder subsidy with the headline "incentives are doing the heavy lifting"[5]. The National Association of Home Builders, which lobbies for builders, paired the sales rise with "affordability challenges," a frame that keeps the conversation on interest rates rather than on builder pricing decisions[7]. Purely market-facing outlets like FXStreet and RTTNews treated the number as a straightforward beat against forecasts, with no household or policy angle at all[8][9]. RISMedia's headline called the gain "slow," then highlighted an 84.9% regional swing in the Midwest, a number drawn from a small enough sample that it moves around a lot from month to month[10].
What's missing from nearly every version is the Census Bureau's own admission that the monthly number could be off by nearly 20 points in either direction[3]. That detail doesn't fit neatly into either the "boom" framing or the "propped up" framing, which may be exactly why it got left out of both.
Summary
Americans bought new single-family homes at a seasonally adjusted annual rate of 684,000 in August 2026. That is according to estimates the Census Bureau and the Department of Housing and Urban Development (HUD) released on September 24[1][2]. The rate was 6.4% above July's revised 643,000. It was the fastest pace since December 2025, an eight-month high[2][4]. Economists had expected about 615,000 to 620,000[2][8][9]. Sales were still 2.0% below August 2025[1].
The headline gain is less certain than it looks. The number comes from a survey, and Census reports a margin of error of ±19.5% on the monthly change. That means the true change could have been a sizable drop or a much bigger jump[3]. The previous month was also revised a lot: July was first reported at 607,000 and is now 643,000[3][8]. Part of August's 'beat' over forecasts comes from that higher starting point.
The main dispute is over what is driving sales. Builders cut prices, and the median new-home price was $393,700. That is 0.4% above July but 5.8% below a year earlier[1][6]. Reuters reported that buyers took advantage of price cuts and incentives. It also said mortgage rates were rising, with the average 30-year fixed rate at 6.95%, and that this remained a drag on the market[2]. Trade outlet HousingWire argued that incentives, mainly builder-paid mortgage-rate 'buydowns,' are doing the heavy lifting[5]. On that reading, the sales pace shows how much builders will pay to move homes, not how strong buyer demand is on its own. Others read the beat as a sign that demand holds up when prices come down[4].
The Event
On September 24, 2026, the Census Bureau and HUD reported that sales of new single-family houses in August ran at a seasonally adjusted annual rate of 684,000[1][3]. That is 6.4% above the revised July rate of 643,000 and 2.0% below the August 2025 rate of 698,000[1]. The median sales price was $393,700[1]. There were 483,000 new homes for sale at the end of August, equal to 8.5 months of supply[1][6].
Undisputed Facts
- August 2026 new single-family home sales ran at a seasonally adjusted annual rate of 684,000, according to Census and HUD[1][3].
- The 6.4% monthly gain carries a Census confidence interval of ±19.5%. The 2.0% year-over-year decline carries one of ±15.7%[3].
- July's rate was revised up to 643,000 from the 607,000 first reported[3][8].
- Forecasters surveyed by Reuters expected 615,000. The Bloomberg survey median was 616,000[2][4].
- The median new-home price was $393,700. That is up 0.4% from July and down 5.8% from August 2025[1][6].
- Of the 483,000 homes for sale, 113,000 were already finished and ready to move into[6].
- The average 30-year fixed mortgage rate was 6.95% in mid-September, the highest since January 2025. Reuters reported rates are up nearly 100 basis points (about one percentage point) since the U.S.-Israeli war with Iran began at the end of February[2][7].
- January 2026 sales were first reported at 587,000, the slowest pace since 2022. The strongest recent month was December 2025[11][12].
The Pressure
Strip away the moralizing and blame. What structural realities persist regardless of which narrative wins?
- Builders must sell what they have built
- Builders carry 483,000 homes for sale, and 112,000 are finished[6]. A finished, unsold home costs money every month in loan interest, taxes and upkeep. So builders have a strong reason to cut prices or pay for buydowns, even at the cost of margin[5][6].
- War-driven borrowing costs
- Reuters reports the Iran war raised energy prices and long-term Treasury yields. It says mortgage rates rose nearly a full percentage point to 6.95%, the highest since January 2025[2]. On a $393,700 home, a point of interest adds a few hundred dollars to a monthly payment. That is roughly what a buydown is designed to offset.
- A survey, not a census
- New-home sales come from a sample of builder contracts. The ±19.5% margin on the monthly change and the 36,000-unit revision to July show how much any one month can move[3]. Beat-or-miss headlines rest on the point estimate alone.
Material realitySeveral facts hold whichever story wins. New-home sales in August were slightly below a year earlier. They were within statistical noise of flat to up from July[1][3]. Prices are falling on a yearly basis: the median is down 5.8%[1]. But the median can also fall if the mix of homes sold shifts toward smaller or cheaper ones. At 8.5 months of supply, the inventory would take most of a year to sell at the current pace. By common rule of thumb, about six months is a balanced market[1][6]. And builders, not buyers, are absorbing much of the higher cost of borrowing through buydowns[5].
Narrative as a weaponMarket-facing outlets (FXStreet, RTTNews, and to a degree Reuters and Bloomberg) shape the first impression with 'beat' and 'eight-month high.' That invites readers to see strength. The home-builder lobby wants readers to see demand held back by interest rates, which is an argument for rate relief and lower costs. Trade press focused on affordability wants readers to see a market kept afloat by builder subsidies. The one thing nearly every headline left out is the Census Bureau's own margin of error. By the agency's own standard, it makes the monthly 'jump' a statistical toss-up. The assigned story's 'highest since January' peg is also wrong. January 2026 was the year's low, at 587,000 as first reported, so the correct comparison is the highest since December 2025[2][4][11].
How Each Side Sees It
Each major actor’s view — how it frames things, its underlying incentive, and how it’s materially affected. Tap a side to read it.
Frames it asThe agencies publish an estimate, not a count. They report every monthly change with a confidence interval. This month that range (±19.5%) is three times the size of the change itself (6.4%)[3]. They revise earlier months as more contract data comes in, which is why July moved from 607,000 to 643,000[3]. From the agencies' view, the honest reading is that sales were roughly flat to up, and below a year ago.
WhyTo protect the credibility of a closely watched indicator by publishing its uncertainty openly and revising as better data arrives[1][3].
Impact on themThe agencies are not affected financially. But their figures move bond and currency markets on release day, and traders react to the single-point estimate[9].
Frames it asBuilders argue that buyers still want new homes and will buy when the monthly payment is reachable. The August beat shows that price cuts and incentives work[2][7]. In their view, the obstacle is borrowing costs, not demand. Builders cannot control rates, which rose after the Iran war pushed up energy prices and Treasury yields[2]. So they absorb the cost themselves. NAHB's own survey found builders' sales expectations for the next six months got much worse in September[2].
WhyTo keep selling a large stock of homes, including 112,000 already finished, without cutting list prices so far that it hurts future pricing[6]. Builders also want lower rates and policies that ease costs[7].
Impact on themBuydowns and price cuts come out of builders' profit margins. HousingWire noted the open question is whether builders still have enough margin to keep sales from falling below 2022 lows if rates keep rising[5].
Frames it asThis camp says the sales figure hides how hard it is to buy. A mortgage-rate buydown is when the builder pays the lender a lump sum upfront so the buyer gets a lower interest rate, either for the whole loan or for the first few years. The buyer sees a smaller monthly payment. But the builder is really cutting the price in a form that does not show up in the median. HousingWire cited reports that 80 to 90% of new-home sales now need buydowns[5]. It also cited Lennar's report that about half of the prospective buyers visiting its communities could not qualify for a mortgage, even though its average price was down 3%[5]. On this view, the market is being held up by subsidy, not strong demand.
WhyTo keep attention on whether ordinary households can afford a home, not on whether one monthly number beat a forecast[5][6].
Impact on themBuyers who do buy get lower prices and subsidized rates. Many others are still shut out by rates near 7% and tighter lending standards[2][5].
Frames it asTraders treat the release as a surprise against expectations. A result of 684,000 against about 615,000 to 620,000 is read as a strong upside beat[8][9]. Bloomberg framed it as the fastest pace this year, reached as prices drop[4]. Market analysts also see housing as the part of the economy most sensitive to interest rates, which makes this report an early read on how the war-driven rise in rates is spreading[2].
WhyTo price bonds and currencies correctly. A forecast miss this large moves those prices whatever the margin of error[9].
Impact on themBeats and misses can briefly move Treasury yields and the dollar, and those feed back into mortgage rates[2][9].
Like this article?
The Bias Ledger average rating 3
The same story, as framed by outlets across the spectrum, ordered least to most biased. The bias score (1 = straight, 10 = heavily spun) is an AI assessment of that framing — click an outlet to see its track record. The tell is the word choice or omission that reveals the angle.
| Outlet | Vantage | Bias | How they frame it | The tell |
|---|---|---|---|---|
| FXStreet | Spain-based forex/markets data site | 1 | United States New Home Sales (MoM) came in at 0.684M, above expectations (0.62M) in August | A pure data headline for traders. It uses a 620,000 consensus rather than 615,000, a reminder that the size of the 'beat' depends on which survey is cited. |
| Bloomberg | U.S. center, business/markets | 2 | US New-Home Sales Rise to Fastest This Year as Prices Drop | 'Fastest this year' is accurate, and the headline puts the price cuts next to the sales gain. The headline has no year-over-year comparison. |
| RTTNews | U.S.-based financial newswire | 2 | U.S. New Home Sales Rebound More Than Expected To Eight-Month High In August | A straight forecast-versus-actual frame. It notes the forecast was based on the unrevised 607,000 July figure. |
| Reuters | U.S./international center wire | 3 | US new home sales jump to eight-month high in August | 'Jump' and 'surged' lead the story. But the body credits price cuts and incentives, calls rising mortgage rates a drag, and gives regional detail. It does not report the ±19.5% margin of error. |
| HousingWire | U.S. mortgage/real-estate trade press | 4 | New home sales are holding up, incentives are doing the heavy lifting | Frames the gain as bought with builder subsidies. It leans on the 80 to 90% buydown figure and Lennar's qualification data. Those points are real, but the frame is skeptical from the start. |
| National Association of Home Builders | U.S. homebuilder industry lobby | 4 | New Home Sales Rise as Affordability Challenges Continue | Pairs the gain with 'affordability challenges' and high rates. That keeps attention on borrowing costs, the issue builders lobby on, rather than on builder pricing. |
| RISMedia | U.S. real-estate brokerage trade press | 5 | New-Home Sales Slowly Rise; Midwest Grows Nearly 85% | 'Slowly' undersells a 6.4% gain. Then the headline highlights the noisiest number in the report, an 84.9% regional swing drawn from a small sample. |
References
- Monthly New Residential Sales, August 2026 — U.S. Census Bureau / HUD · U.S. federal statistical agency (primary data)
- US New Home Sales Jump to Eight-Month High in August — Reuters (via U.S. News & World Report) · International wire service, center
- New Home Sales Increase to 684,000 Annual Rate in August — Calculated Risk · Independent economics newsletter (Bill McBride), data-focused, reader-funded
- US New-Home Sales Rise to Fastest This Year as Prices Drop — Bloomberg · U.S. business/markets news, center; owned by Bloomberg L.P., a financial data firm
- New home sales are holding up, incentives are doing the heavy lifting — HousingWire · Mortgage and real-estate industry trade publication, advertiser-funded
- Builders cut prices as new home supply holds at 8.5 months — HousingWire · Mortgage and real-estate industry trade publication, advertiser-funded
- New Home Sales Rise as Affordability Challenges Continue — NAHB Eye on Housing · Blog of the National Association of Home Builders, the builder industry's lobby
- U.S. New Home Sales Rebound More Than Expected To Eight-Month High In August — RTTNews · Commercial financial newswire
- United States New Home Sales (MoM) came in at 0.684M, above expectations (0.62M) in August — FXStreet · Spain-based commercial forex/markets data site
- New-Home Sales Slowly Rise; Midwest Grows Nearly 85% — RISMedia · Real-estate brokerage industry trade publication
- HUD and Census Bureau Report New Residential Sales in January 2026 — U.S. Department of Housing and Urban Development · U.S. federal agency (primary data)
- January new home sales plunge to the slowest pace since 2022 — CNBC · U.S. business news, center; owned by Comcast/NBCUniversal