Skip to main content
Education only. ClearMoneySchool does not provide individualized investment, tax, or legal advice. Why we don't give advice →

A Monthly Report On New Homes Leads With A Figure Near 1.3 Million. About 110 Thousand Homes Were Actually Begun In The Month It Describes.

The bigger figure is what a full year would come to if that pace held and if no stretch of the calendar were busier than another. It is a level put on an annual footing, not a count of what happened. The same document prints both, a few pages apart, and only one of them gets quoted.

EconomyNovember 18, 2026Edited by Joseph Citizen, FounderUpdated November 18, 2026
Download MP3
0:000:00

The simple version

A monthly report on new homes leads with a figure near 1.3 million, and about 110 thousand homes were actually begun in the month it describes.

The bigger figure is what a full year would come to if that pace held and if no stretch of the calendar were busier than another. It is a level put on an annual footing.

The smaller figure is the count. Both sit in the same document a few pages apart, and only one of them gets quoted.

The numbers

  • The report is Monthly New Residential Construction, October 2026, release number CB26-187, issued jointly by the U.S. Census Bureau and the Department of Housing and Urban Development. It carries the line FOR RELEASE AT 8:30 AM EST, WEDNESDAY, NOVEMBER 18, 2026. Read November 18, 2026.
  • The quoted figure. Privately-owned housing starts in October were at a seasonally adjusted annual rate of 1,309,000. The release labels the three headline figures Seasonally Adjusted Annual Rate (SAAR) directly beneath them.
  • The count. In the same release, Table 3b, headed Not seasonally adjusted, gives October 2026 starts as 109.7 thousand, marked preliminary. Of that, 71.1 thousand were single-family and 37.6 thousand were in buildings with five units or more.
  • So the figure in the headline is about 11.9 times the number of homes the same document says were begun. Neither number is wrong and neither is hidden. They answer different questions.
  • What the issuer says the term means. The Census Bureau's Survey of Construction Seasonal Adjustment FAQs answers the question What is an annual rate? directly: the seasonally adjusted annual rate for a month is an estimate of what the annual total would be if non-seasonal conditions were the same all year, and it states that this rate is not a rate in a technical sense but is a level estimate.
  • The same document gives the arithmetic as the seasonally adjusted monthly value taken twelve times, and sets it out as the unadjusted monthly survey estimate, set against the seasonal factor, scaled to twelve. It also defines seasonal adjustment as estimating and removing seasonal effects from a time series.
  • Why the practice exists, in the issuer's words: the benefit of the annual rate is that one month's data or one quarter's data can be set against an annual total, and a month can be set against a quarter. The same FAQ notes that the Bureau of Economic Analysis publishes quarterly gross domestic product on the same footing.
  • The other two headline figures, for completeness. Building permits were at a seasonally adjusted annual rate of 1,295,000, which the release puts 7.4 percent below the revised September rate of 1,398,000 and 8.7 percent below the October 2025 rate of 1,418,000. Completions were at 1,272,000.
  • One caution the release prints itself. It states that month-to-month changes in seasonally adjusted statistics often show movements which may be irregular, and that it may take three months to establish an underlying trend for building permit authorizations and six months for total starts and total completions. The next release is scheduled for December 17, 2026.
  • The starts comparisons also carry a margin of error and an asterisk, which we covered separately in our piece on the housing starts margin of error. That is a different point from this one and it is not repeated here.
  • For the other kind of annualizing, where a monthly percentage change is compounded across a year rather than a level being scaled, see our piece on annualized inflation. The two use different arithmetic and are easy to confuse.

Why a month gets put on an annual footing at all

A raw monthly count cannot be set against a full year without being scaled first, and it cannot be set against another month without the calendar being taken out of it. Both problems get solved in the same step.

The scaling answers the comparison problem. A figure stated on an annual footing can be held against last year's total or against a quarter, which a bare monthly count cannot.

The seasonal step answers the calendar problem. New work gets under way more often in some parts of the year than others for reasons that have nothing to do with demand, so each month is weighed against how busy it normally runs before anything is scaled.

The cost of doing both at once is that the published figure no longer looks like anything you could count. It is an estimate of a pace, expressed in the units of a year, and the count it came from sits in a table further back.

The actual math on 1,309,000 and 109,700

The issuer prints the recipe. Start with the raw count for the month, set it against the factor that says how busy that month normally runs, and scale what is left to twelve months.

Run it in reverse and the factor falls out. October's raw count of 109.7 thousand, scaled to twelve months with nothing taken out for the calendar, comes to 1,316.4 thousand. The published figure is 1,309,000.

Set one against the other and the implied factor is about 1.006, which says October ran almost exactly level with an average month. That is why the two figures land so close together here, and it will not be true of a month that is genuinely busy or genuinely quiet.

One more comparison worth running. The raw count went from 101.1 thousand in September to 109.7 thousand in October, a rise of 8.5 percent, while the release states the adjusted rise as 9.9 percent. Both describe the same month and the gap between them is the seasonal step.

The scaling to 1,316.4, the implied factor of about 1.006, the 11.9 times comparison and the 8.5 percent raw change are ours. Every figure going into them is the release's own.

What this means

When a figure is described as an annual rate, the useful question is what the underlying count was. The raw number is usually published in the same document, and it is the one that tells you how much actually happened.

The second habit is to notice which of two adjustments is doing the work. A figure can move because the thing it measures moved, or because the correction for the time of year moved, and the two are separable when both tables are printed.

Plain-English takeaway

This article explains what an annual rate is in one statistical release and reports one month of its figures. It is general education, not investment advice, and it is not a forecast of any future figure.

What this is NOT

This article explains that the headline figures in the Census Bureau and HUD release on new residential construction are levels scaled to twelve months and corrected for the time of year, rather than counts of homes begun, using the October 2026 release read on November 18, 2026 and the Census Bureau's own Survey of Construction Seasonal Adjustment FAQs as its sources. It is not a claim that any figure in the release is wrong or misleading: the Census Bureau publishes the unadjusted count in the same document and explains the method in its own FAQ, which is the opposite of concealing it. It is not a forecast of housing construction, home prices, rents or mortgage rates, and not advice about whether to buy, sell, build or rent. It is not advice about any security, fund or sector, including any homebuilder. It is not a statement about the margin of error on these estimates or about whether any change in them is statistically significant, which is a separate matter covered in our earlier piece and not repeated here. It is not a description of how any other statistical agency annualizes any other series: only this release and this issuer's FAQ were read, and the compounding method used for monthly percentage changes is a different calculation. Ours are the scaling of the raw count to 1,316.4 thousand, the implied seasonal factor of about 1.006, the statement that the headline is about 11.9 times the raw count, and the 8.5 percent raw monthly change; every underlying figure is the issuer's. This is not a political endorsement or criticism of anyone.

Sources

  • U.S. Census Bureau and U.S. Department of Housing and Urban Development, Monthly New Residential Construction, October 2026, release number CB26-187, at census.gov/construction/nrc/pdf/newresconst_202610.pdf, read November 18, 2026, for the release line and date, the 1,309,000 starts figure and its SAAR label, the 1,295,000 permits and 1,272,000 completions figures and their comparisons, the Table 3b unadjusted October count of 109.7 thousand and its single-family and five-unit detail, the September unadjusted count of 101.1 thousand, the explanatory note on irregular movements and trend periods, and the December 17, 2026 next release date.
  • U.S. Census Bureau, Survey of Construction Seasonal Adjustment FAQs, at census.gov/construction/pdf/soc_seas_faqs.pdf, read November 18, 2026, for the definition of seasonal adjustment, the answer to What is an annual rate?, the statement that the rate is a level estimate rather than a rate in a technical sense, the arithmetic of the monthly value taken twelve times, and the note on quarterly gross domestic product.
Receipts2 receipts: the source's own words, with a link to eachShow
  1. an estimate of what the annual total would be if non-seasonal conditions were the same all year
    U.S. Census Bureau, Survey of Construction Seasonal Adjustment FAQs, question 13Read Nov 18, 2026
    Open the source (opens in a new tab)
  2. Privately-owned housing starts in October were at a seasonally adjusted annual rate of 1,309,000
    U.S. Census Bureau and HUD, Monthly New Residential Construction, October 2026, release CB26-187Read Nov 18, 2026
    Open the source (opens in a new tab)

Found this useful?

Education only. Nothing here is investment, tax, or legal advice.