ADU statistics

Production: permits and completions, 2018–2025

Permits and completions are the backbone series for the entire category. California jurisdictions report building permits and certificates of occupancy to HCD under the Housing Element annual reporting requirement, and Table A2 carries a UNIT_CAT field with an explicit ADU value.

The permit and completion columns are agency-published counts of unique events, deduplicated by tracking identifier; the ratio columns are ours. Full method and the four unusable reporters on the permits-vs-completions study.

Permit and completion counts are agency-published. Every ratio column is a HyreADU calculation from those two counts. Statewide totals: 178,495 permit units, 102,356 completions, 57.3% cumulative. Excluding the four jurisdictions whose completion fields are unusable — Huntington Beach, Saratoga, West Sacramento and Arvin — the figures are 177,426, 102,352 and 57.7%. Retrieved 2026-09-05.

A naive sum by APR reporting year rather than by the date on the permit double-counts 35,237 permit units, 16.5% of a naive 213,732-unit total. Roundups that report "over 200,000 California ADU permits" are usually reporting that artefact.

Permit cohorts followed to completion

Each row follows the ADU permit units dated to one year and asks how many have a matching completion row today. Follow-up length differs by row, so the rows are not comparable to one another as performance measures — only as a maturing series.

HyreADU calculation on HCD APR Table A2, retrieved 2026-09-05. Matching is on the jurisdiction’s own project identifier; 15,082 ADU rows (5.4%) carry no identifier at all and cannot be matched in either direction. The 2025 row has zero years of follow-up and is printed only so the series is complete.

Full method, the identifier-join limitations and the treatment of the four unusable reporters are on the permits-vs-completions study.

Where the permits are: the twenty largest reporters

ADU production in California is concentrated. The City of Los Angeles alone accounts for 27.8% of every ADU permit unit reported statewide since 2018, and the twenty jurisdictions below account for 53.7% of them between them.

The completion ratio column is a reporting-quality signal as much as a construction signal, and this table is emphatically not a ranking.

HCD APR Table A2, 2018–2025, retrieved 2026-09-05. Counts are agency-published; the ratio is ours. Four jurisdictions are excluded from every table on this page as unusable reporters — Huntington Beach (495 ADU permits, zero completions ever reported), Saratoga, West Sacramento and Arvin. They are named rather than silently counted as zero production.

A low ratio can mean slow construction, a reporting practice that never fills the completion field, or an identifier that changed. It cannot be read as a jurisdiction’s competence. The permits-vs-completions study sets out the four rules used to classify a reporter as unusable.

The two clocks: approval, then construction

"How long does an ADU take" is two questions with two different datasets behind them. The first clock runs from application to permit issuance and is measurable in three California jurisdictions that publish both dates.

The second runs from issuance to certificate of occupancy and is measurable in exactly one, Los Angeles, which publishes an issue date and a certificate date on the same permit row.

Medians throughout, never means, and the interquartile range beside every median because these distributions are strongly right-skewed.

HyreADU calculations on municipal open data, retrieved 2026-09-05. The issuance-to-certificate median is a Kaplan-Meier estimate that keeps the 14,306 permits which have not completed inside the calculation; the finished-permits-only row is printed beside it because it is the number a builder’s own records would produce, and it is roughly four months shorter for a reason that has nothing to do with building faster.

Neither clock is the statutory 60 days. Government Code § 66317 runs 60 days from a completed application, and none of these extracts marks the completeness determination, so applicant delay, fee payment and back-check cycles all sit inside the median.

A median above 60 days is not by itself a finding that a city missed the shot clock. Full treatment on the permit-times study and the construction-duration study.

Construction time by what is actually being built

The permit type is the closest thing the Los Angeles data has to a construction method. A new-construction permit is a detached ADU built from the ground up; an alteration permit is usually a garage or interior conversion. The gap between them is five months at the median.

HyreADU calculation on LADBS resource pi9x-tg5x, retrieved 2026-09-05, observation window closing 2026-08-31. Kaplan-Meier medians. City of Los Angeles only.

Junior ADUs finish fastest of all — median 387 days against 509 days for a standard ADU — which is what you would expect of a unit created inside an existing bedroom footprint. See the construction-duration study.

Size: the statute is choosing the number

The floor-area distribution of new detached ADU permits in Los Angeles is not shaped like a distribution of household needs.

It has hard modes sitting on the numbers written into the Government Code, with a cliff immediately above each one. 635 permits declare exactly 1,200 square feet — the statutory ceiling for a detached ADU under Gov. Code § 66314(a)(5) — and 1,061 sit in the last five square feet below it. Only 12.5% go above.

HyreADU calculation on 8,536 City of Los Angeles new-construction ADU permits with a declared floor area between 100 and 3,000 sq ft, issued 2020 to 2026-08-30. LADBS resource pi9x-tg5x, retrieved 2026-09-05. Floor area is the department’s field, not a HyreADU measurement.

The policy reading is available and worth stating carefully: a size cap set at 1,200 square feet does not merely constrain the largest units, it creates a cluster at 1,200.

The size-distribution study shows the same shape at 800 and at 750, and treats the two recodifications of the ADU statute since 2024 explicitly — § 65852.2 is stale and should not be cited.

Declared permit valuation, five jurisdictions

Declared permit valuation is the closest thing to a cost figure that exists in public permit data, and it is not a cost figure.

Declared valuation is the value an applicant declared to the building department, typically because fees are calculated from it.

The spread between jurisdictions is large enough that a statewide average would be meaningless, which is why we do not publish one.

HyreADU calculation on municipal permit harvests taken 29–30 August 2026, computed 2026-09-05. Owner-builder records are excluded. Issued-date windows differ by city and are stated in the ledger. Zero and missing valuations are dropped.

Within Los Angeles the split by ADU type is the more useful cut: garage conversions have a median declared valuation of $27,000 against $80,000 for everything else — a 2.96× difference on the same permit form, in the same city, in the same years.

Declared valuation is not construction cost; see the permit-valuation study for what it omits.

How much of California’s housing is now accessory

Statewide, ADUs went from 9.1% of HCD-reported housing permits in 2018 to 26.5% in 2024. Underneath that state average the jurisdictional spread is roughly ten to one.

The two share columns below answer different questions and are never mixed: one is a share of what a city has permitted, the other a share of what the state told it to plan for.

HyreADU join of HCD APR Table A2 to the state’s 6th-cycle RHNA Progress Report, both retrieved 2026-09-05. Cycle start dates run from April 2021 to December 2023 depending on the region, so the two windows do not align exactly and the shares are approximations, labeled as such.

Statewide, 477,627 units of all types have been permitted against an allocation of 2,495,457 across 539 jurisdictions — 19.1%.

An allocation is a planning obligation, not a construction target, and no ADU on this page can be assigned to an income band because Table A2’s ADU rows carry no affordability field.

The production-vs-need study is explicit about that limit; it is the single most consequential error available on this topic.

Every figure on this page, with its class

The audit table. 68 figures, each with its geography, its date, its provenance class and its source.

This is the list the one-third gate counts: 48 HyreADU calculations, 16 agency-published, 4 trade or survey.

If a number appears anywhere above and is not in this table, that is a defect and we would like to hear about it.

Class definitions: HyreADU calculation — HyreADU computed it from a primary file, method on the linked study. Agency published — an agency printed it. Trade / survey estimate — a named survey or industry estimate. All retrieval dates as stated; the HCD, LADBS, DataSF, Marin and RHNA extracts were all taken on 2026-09-05.