Washington's annual new housing authorization totals fell sharply after 2022 in the Census Bureau's Building Permits Survey. The District recorded 7,705 authorized housing units in 2022 and 1,591 in 2025 in the annual county files reviewed for this article. That is a change in a defined construction indicator. It is not a count of apartments currently available to rent or a direct forecast of what rents will do next.
The most useful reading combines the total with the types of structures behind it. In every year from 2021 through 2025, most authorized units were in buildings containing five or more units. That concentration matters when interpreting annual swings: a relatively small number of large projects can contribute many units, while the count of buildings tells a different story.
The five year record for the District
The figures below come from the Census Bureau's annual county files, selecting state code 11 and county equivalent code 001 for the District of Columbia. Housing units were summed across the one unit, two unit, three and four unit, and five or more unit structure categories. These are annual authorization totals, not seasonally adjusted monthly rates.
In 2021, the District total was 4,740 units. Of those, 376 were in one unit structures, 24 in two unit structures, seven in structures containing three or four units, and 4,333 in structures containing five or more units. The largest structure category therefore accounted for about 91.4 percent of the total.
In 2022, the total rose to 7,705 units, including 409 in one unit structures, 64 in two unit structures, four in the three and four unit category, and 7,228 in the five or more unit category. The last category represented about 93.8 percent of that year's total.
In 2023, the total was 3,020 units, with 166 in one unit structures, 64 in two unit structures, eleven in the three and four unit category, and 2,779 in the five or more unit category. In 2024, the corresponding total was 1,737, comprising 146, 82, three, and 1,506 units respectively.
In 2025, the total was 1,591 units: 181 in one unit structures, 38 in two unit structures, none in the three and four unit category, and 1,372 in buildings containing five or more units. That last category contributed about 86.2 percent of the year's total. The mix remained strongly concentrated in larger residential structures despite the lower overall total.
| Year | One unit | Two units | Three or four units | Five or more units | Total units |
|---|---|---|---|---|---|
| 2021 | 376 | 24 | 7 | 4,333 | 4,740 |
| 2022 | 409 | 64 | 4 | 7,228 | 7,705 |
| 2023 | 166 | 64 | 11 | 2,779 | 3,020 |
| 2024 | 146 | 82 | 3 | 1,506 | 1,737 |
| 2025 | 181 | 38 | 0 | 1,372 | 1,591 |
Compare the changes without hiding the starting point
The increase from 4,740 units in 2021 to 7,705 in 2022 was 2,965 units, or about 62.6 percent. The following year's decrease was 4,685 units, or about 60.8 percent. A reader who sees only one of those percentages misses the unusually high 2022 comparison point.
The total then decreased by 1,283 units from 2023 to 2024, about 42.5 percent, and by 146 units from 2024 to 2025, about 8.4 percent. The smaller final percentage decline means the annual total changed less between those last two years. It does not mean the total had returned to its earlier level.
Across the entire five year period, the annual files sum to 18,793 authorized units. The 2025 figure is about 66.4 percent below the 2021 figure and about 79.4 percent below the 2022 peak. These calculations describe this selected period and geography; choosing another starting year would answer a different historical question.
The five or more unit category fell by 5,856 units between 2022 and 2025. The other categories together fell from 477 to 219 units, a decrease of 258. Thus most of the numerical decline between those years occurred in the largest structure category. This is an arithmetic decomposition, not an explanation of the economic causes.
Keep buildings and housing units separate
A building containing many apartments contributes one building and many housing units. The Census files contain both kinds of fields. For a housing supply discussion, copying the building count into a column labeled units would substantially misstate the scale of larger projects. This article uses the housing unit fields consistently.
The distinction also changes how a reader thinks about volatility. One large authorization can noticeably alter an annual unit total without implying that hundreds of independent projects began at once. The structure mix shows where to ask more questions, but it does not identify the individual projects responsible for a particular year.
A two unit building contributes two units to the relevant category. The three and four unit category combines structures of both sizes, so its unit total need not be a multiple of one single building size. Keeping the category definitions visible helps avoid treating a small or unusual total as an error without checking the source.
For a neighborhood question, the District wide table is only a starting point. It does not show which ward, corridor, or block contains the authorized units. A household considering a particular location needs project level information and current property availability, not an assumption that the citywide total is distributed evenly.
Authorization is one stage of development
The Building Permits Survey measures housing units authorized by building permits within its defined scope. Census distinguishes authorization from starts, construction in progress, and completion. A unit can appear in an authorization year without becoming available for occupancy in that same year.
Consider a fictional project authorized late in one year, started in another, and completed later. Its units belong to the authorization total for the first event, while its effect on available homes occurs on a different schedule. This invented timeline illustrates why a low authorization year can coexist with openings from projects approved earlier.
The annual table cannot establish how many authorized projects were delayed, changed, or completed. Nor does it identify which units were offered for rent rather than sale. A statement that Washington added 1,591 available rentals in 2025 would therefore go beyond the figures presented here.
When discussing a specific project, use the relevant local permit, construction, and occupancy records and the property's current information. Keep each milestone dated. A public authorization statistic and a leasing advertisement can both be accurate while describing different stages and different sets of homes.
Understand what the series leaves out
Census describes the series as new privately owned residential construction. Its definitions exclude certain categories, including units created within an existing residential or nonresidential structure. That limitation is especially relevant when a local discussion concerns converting existing buildings rather than constructing new ones.
The series is also not a net change in the housing stock. A net measure would need to account for losses and other changes outside these authorization totals. Do not subtract one year's permits from another and describe the result as homes removed from the District. The subtraction shows a change in annual authorizations.
The data do not directly state asking rents, vacancy, household incomes, affordability restrictions, or the condition of existing apartments. Those topics require additional evidence. A strong decline in authorizations may motivate a question about future supply, but the table alone cannot determine the size or timing of a rental market effect.
Census notes that annual data can include estimates with imputation and that its permit office universe methodology changed beginning in 2023. Readers making a formal statistical comparison should consult the survey documentation and any relevant coverage notes. The practical lesson is to retain the source definition and release version with the numbers.
Reproduce the table before using it elsewhere
Download each annual county file from the Census source links below and select the District's state and county equivalent codes. Confirm the year and geography in the row. Then identify the housing unit columns for each structure category rather than the adjacent building or valuation columns.
Sum the four housing unit categories and compare the result with the totals stated here. Keep the original row and the calculation in the working file. This makes it possible to investigate a discrepancy without relying on a copied chart whose source fields are no longer visible.
For percentage change, subtract the earlier total from the later total and divide by the earlier total. Multiply by 100 and state whether the result is an increase or decrease. Rounding to one decimal place is adequate for the comparisons in this article and avoids implying more interpretive precision than the annual figures provide.
Census announced final annual 2025 estimates on May 14, 2026. Preserve that release context and the date of retrieval if publishing a new chart. If a later file differs, compare versions and documentation before assuming one transcription is necessarily correct. A transparent source trail is part of the analysis.
Use the finding for the right question
For a renter, this series provides background on the pace and composition of new housing authorizations. It should not replace a current search for available units, verified lease terms, or an assessment of a particular property. The annual total cannot tell a household what will be available on its move date.
For a local discussion, the clearest finding is that authorizations declined after a high 2022 total and remained dominated by buildings with five or more units. Further questions should address project timing, location, conversions, completions, and affordability using suitable additional sources. That approach preserves the value of the Census data without asking one indicator to describe the entire Washington housing market.