The Census Bureau reported that new-home sales fell in April. Printed beside that figure, in the same release, is a margin of error twice the size of the fall. Chicago’s own records are counts rather than estimates, and through April they rose.
Cook County recorded sales and City of Chicago permits · January through April, compared like for likeStart with the two national figures. On May 28 the Census Bureau and the Department of Housing and Urban Development reported that sales of new single-family houses ran in April at a seasonally adjusted annual rate of 622,000. They put that at 6.2 percent below March and 11.3 percent below April of the previous year. Separately, in July, the Housing Market Index, a survey in which homebuilders are asked to rate conditions in their own market, read 34, two points below its June reading.
Both are real numbers and both are national. It is worth being exact about what the first one counts, which is newly built single-family houses. The Chicago figures in this article cover houses, condominiums and two- to six-flat buildings of every age. Those are not the same universe, and nothing here treats one as a check on the other.
Both of those percentages carry an asterisk in the release itself. The decline from March is 6.2 percent give or take 12.8. The decline against the previous April is 11.3 percent give or take 11.5. Each margin is wider than the figure it qualifies. Census sets out what that means further down the same document: where the range includes zero, the change is not statistically significant, and it is uncertain whether there was an increase or a decrease. New-home sales are estimated from a sample rather than counted, and for April the sample was not able to establish which way they went.
Chicago’s records are built the other way round. A recorded deed is not a sampled household and an issued permit is not a polled builder. These files set out to record transactions as they happen rather than to estimate them from a sample, so within whatever we ask them for there is no sampling margin to report. That does not make them flawless. Administrative files have their own failure modes: a sale can be coded to the wrong property class, a record can be entered wrong, and we lean on the Assessor’s class codes to keep houses, condos and flats apart without having audited them. But the failure mode that nearly cost us this article is completeness, meaning simply whether everything that happened has been written down yet.
What Chicago’s records actually show
Across the five years on record here, all three property types sat lowest in 2023 and have climbed since. The chart shows each one’s 2023 floor and where it stands in 2026, on identical January-through-April windows.
| Median sale price | 2022 | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|---|
| Houses | $300k | $269k | $298k | $310k | $325k |
| Condos | $300k | $293k | $315k | $325k | $345k |
| Two- to six-flats | $379k | $358k | $395k | $402k | $460k |
Houses are up 21% on their 2023 low, condos 18%, and two- to six-flat buildings 29%. That last one is the number worth pausing on. Of the three types measured here, the small multi-unit buildings moved the furthest. The county file records the building, not who lives in it, so nothing here says whether those units are rented or owner-occupied.
And the city is permitting more, not less
| Year | 2022 | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|---|
| New-construction permits | 414 | 332 | 295 | 374 | 439 |
| Arms-length sales | 11,438 | 7,551 | 7,562 | 5,492 | 6,690 |
New-construction permits are at 439 for the first four months of this year, the highest of the five years and +49% above the 2024 trough. Builder sentiment is a national survey of how builders feel and a permit count is a local record of what the city issued, so the two are not the same instrument. What can be said is that more new-construction permits were issued in Chicago in those four months than in the same four months of any of the four years before it. Sales volume tells a more complicated story: it fell hard between 2022 and 2025, from 11,438 to 5,492, and has turned back up to 6,690. Fewer homes are changing hands than four years ago, and more are changing hands than last year.
The number we nearly printed, and why we did not
This piece was very close to saying Chicago sales had collapsed by 59 percent. That figure was an artifact and it is worth explaining, because anyone running the same query will hit it. Cook County records deeds on a delay, and the delay is not a gentle taper. When we retrieved the file on July 19, 2026, it held no June or July sales at all, and only a part-month for May.
| Recorded sales | 01 | 02 | 03 | 04 | 05 | 06 | 07 |
|---|---|---|---|---|---|---|---|
| 2025 | 1,016 | 1,246 | 1,041 | 2,189 | 2,389 | 1,672 | 2,020 |
| 2026 | 1,258 | 1,426 | 1,971 | 2,035 | 969 | 0 | 0 |
Compare January to June across years on that file and you set four and a half months of 2026 against six months of everything else. It manufactures a collapse. Cut every year to January through April, the last month whose count does not obviously truncate, and the same comparison runs the other way: 5,492 sales in 2025 against 6,690 in 2026. Every Chicago figure on this page uses that January-to-April window. The monthly table just above is the exception, and deliberately so, because its whole purpose is to show where the record stops.
What this does and does not establish
- On the recordChicago median recorded sale prices are above their 2023 low in all three property types: houses +21%, condos +18%, flats +29%.
- On the recordNew-construction permits for January through April are at 439, the highest of the last five years.
- On the recordThe mix between the three types barely moved: houses ran 39-41 percent of sales across all five years, so the rise is not simply cheaper homes dropping out of the sample. This does not rule out composition shifts inside a type.
- On the recordArms-length sales for January through April rose from 5,492 in 2025 to 6,690 in 2026, while remaining well below the 11,438 of 2022.
- Not establishedThat Chicago contradicts the national figures. It does not, because they are not the same measure. One national number is a sample survey of sales of newly built homes; the other is a survey of how builders rate their own market. These are recorded sale prices of homes of every age, new and existing alike, because the county file carries no flag that reliably separates the two. All of it can be true at once.
- On the recordCensus flags both of its April percentage changes as not statistically significant. That is the agency’s own determination, printed beside the figures, not our reinterpretation of them.
- Not establishedThat the national market is fine. Not being able to measure a decline is not the same as there being no decline. The honest reading is that this particular release cannot settle it.
- Not establishedWhy Chicago’s numbers moved the way they did. Deeds and permits record what happened, not why, and nothing here identifies a cause.
- Not establishedWhat happens next. A market can be recovering in April and turn in September, and this file will not show that for months.
How we counted
- Chicago means the city’s eight assessment townships, codes 70 through 77, which is how the Assessor divides Chicago. Suburban Cook is not in any figure here.
- No margins of error on the Chicago numbers, because they are not estimates from a sample. They are counts of records that met the filters set out here: recorded sales passing the arms-length tests in the three property classes, and permits of the single type PERMIT – NEW CONSTRUCTION. Every other permit type the city issues is outside this article. The risk in an administrative count is not sampling error but incompleteness, which is what the window section is about.
- Sales: Cook County’s recorded sales file, dataset wvhk-k5uv, filtered with the county’s own arms-length flags: no sub-$10,000 transfers, no excluded deed types, no repeat sale inside 365 days, no multi-parcel sales. Two screens are ours rather than the county’s, and we would rather name them than bury them: we drop sales under $20,000 and over $5,000,000, to keep nominal transfers and trophy sales from pulling the median. Every year is filtered identically, so the comparison between years holds either way.
- The window is the method. January through April for every year. April is the last month whose count does not obviously truncate: May is part-recorded and June and July are empty. That is an inference from the shape of the file, not a completion notice, and these counts can still be revised upward as deeds are recorded late. See the section above; a Jan-Jun window reverses the sign on the volume finding.
- Property types are kept apart. Assessor classes 202-210, 234, 278 and 295 are houses; 299 is a condominium; 211 is a two- to six-flat. Combining them lets a shift in the mix move the median with no price change anywhere, so each is reported on its own.
- Permits: City of Chicago building permits, dataset ydr8-5enu, permit type PERMIT – NEW CONSTRUCTION, current through 2026-07-17 at the time of writing. Because the sales file surprised us, we ran the same test here: monthly counts of new-construction permits for January through July, 2025 and 2026. Every month from January to June 2026 came back populated and in line with 2025, running 68, 82, 132, 157, 130 and 129 against 75, 80, 95, 124, 131 and 161. July 2026 stood at 91 against 160, which is what a live file looks like when you read it mid-month: the most recent permit in it had been issued on July 17. So this file shows no hole of the kind the sales file has, and the January-to-April window sits well clear of its edge. It does not follow that the file is final. Records can still be added or amended later, here as in the sales file.
- National figures, with their dates and their publishers: April 2026 new-home sales are a joint series from the U.S. Census Bureau and the U.S. Department of Housing and Urban Development. We read the agencies’ own release, CB26-85, issued May 28, 2026, and both percentage changes we quote carry Census’s asterisk for not statistically significant. Builder confidence is a different kind of number: the NAHB and Wells Fargo Housing Market Index, a survey in which homebuilders rate their own market. It read 34 for July 2026. That page carries only the current month, so for the July reading as reported at the time see this account of it. We have not recomputed either figure, and both are national rather than Chicago.
- Medians, not averages, because a handful of very large sales pull a mean upward and away from what an ordinary transaction looked like.
- Run it yourself. Both datasets answer Socrata queries. Everything above was retrieved on July 19, 2026 with these filters. Sales: township_code in (‘70’…‘77’) and sale_filter_less_than_10k=false and sale_filter_deed_type=false and sale_filter_same_sale_within_365=false and is_multisale=false and num_parcels_sale=1, with sale_date inside January 1 to May 1 of each year, sale_price above 20,000 and below 5,000,000, and the class list for each property type given above. Permits: permit_type = ‘PERMIT – NEW CONSTRUCTION’ with issue_date inside the same January-to-May window. The monthly counts in the table use the identical sales filters, one month at a time, which is why they add up to the yearly totals.
Computed by KCM Desk from Cook County recorded sales and City of Chicago permit records; published July 19, 2026. If you spot an error, corrections come first.