Research data
ADU research data — the files, not the screenshots
Nine CSV files, CC BY 4.0, each one the file a HyreADU study was computed from. Every row count on this page was read from the published file itself.
Written by HyreADU Research Desk Primary-source research and data analysis
Audited by HyreADU Research Desk File inventory and provenance audit
What is here
9 CSV files, 3,828 data rows, 478 KB in total, published under Creative Commons Attribution 4.0. Each one is the underlying file for a HyreADU study, not a summary of it: the same numbers the charts and tables on those pages are drawn from, in the shape they were computed in.
7 of the 9 are HyreADU calculations — a primary file was downloaded and the figure computed here — and 2 are agency-published data this desk compiled rather than produced, labeled as such in their own row so nobody credits us with HUD’s rent survey or HCD’s returns.
The largest is the permit contractor index at 3,668 rows, one row per contractor per jurisdiction across five California permit feeds; the smallest is four rows of HUD Fair Market Rent.
Every file carries a comment line at the top stating what the file is, what it is not, the retrieval date, and the exclusions applied — because the mislabelling risk on this material is real. Declared permit valuation is not construction cost.
A permit is not a completion. Fair Market Rent is not an asking rent.
What is not here is anything we have not computed. There is no statewide fee file, because the city-by-city fee study hit a retrieval wall and we would rather ship nothing than ship a table of aggregator numbers under our name.
Why a research site publishes its files
A statistic you cannot open is a statistic you have to take on trust.
Most ADU numbers in circulation are exactly that: a figure in a headline, a chain of citation behind it, and at the end of the chain a builder’s brochure or a page that has been deleted.
That is the condition this desk was set up to work against, and publishing the file is the only version of the argument that is not itself an assertion.
There is a second reason, which is more selfish. Publishing the file constrains us. A median that has to survive somebody else recomputing it from the same rows is a median we check twice.
An exclusion rule that has to be stated in a header comment is an exclusion rule we have to be able to defend.
Several of the caveats on these files — owner-builder records excluded, zero valuations dropped, negative durations dropped — exist in this form because they had to be written down for a stranger.
And a third: these files are the answer to the question a journalist actually asks. Not "what is your headline number" but "can I see it, can I cut it a different way, and can I say where it came from". All three are yes here, and the license says so in writing.
The files
Click a filename to download. Click "Method" to read the study the file was computed for — the exclusions, the frame and the sensitivities live there, not here.
| File | What it is | Rows | Cols | Size | Provenance | Method |
|---|---|---|---|---|---|---|
| adu-statute-change-log.csv | California ADU statute change register | 36 | 9 | 4.9 KB | HyreADU calculation | Method |
| adu-statute-retired-numbers.csv | Retired California ADU statute section numbers | 4 | 4 | 0.9 KB | HyreADU calculation | Method |
| adu-permit-contractor-index.csv | ADU permit contractor index, five jurisdictions | 3,668 | 14 | 445 KB | HyreADU calculation | Method |
| adu-construction-duration-la.csv | ADU permit-to-occupancy duration, City of Los Angeles | 24 | 10 | 1.8 KB | HyreADU calculation | Method |
| adu-permit-times-by-city.csv | Application-to-issuance permit times, three jurisdictions | 17 | 9 | 3.1 KB | HyreADU calculation | Method |
| adu-declared-permit-valuations.csv | Declared building-permit valuations, five jurisdictions | 20 | 13 | 4.1 KB | HyreADU calculation | Method |
| adu-permits-vs-completions.csv | ADU permits against completions, top 20 usable jurisdictions | 20 | 8 | 1.3 KB | Agency published, compiled by HyreADU | Method |
| adu-preapproved-plan-programmes.csv | Pre-approved ADU plan programs | 35 | 8 | 14.9 KB | HyreADU calculation | Method |
| adu-hud-fmr-fy2026.csv | HUD Fair Market Rents FY2026, four California FMR areas | 4 | 9 | 1.4 KB | Agency published | Method |
Row counts exclude the header row and the comment line each file carries. Sizes are the files as served. Inventory read 2026-09-07.
Every file is UTF-8, comma-separated, with a single leading comment line beginning # and then one header row. If your tool chokes on the comment line, skip the first line.
What each file is, and what it is not
The traps in this material are specific and they are all mislabelling traps. Each file is described here in the terms that stop it being misread.
adu-permit-contractor-index.csv — the largest file
3,668 rows, one per contractor per jurisdiction, across five California permit feeds: the City of Los Angeles, San Francisco, Sacramento, San José and unincorporated Marin.
Carries the name on the permit, the license number recorded on the permit, ADU sites and permits counted, first and last permit date, declared valuation, and the matched CSLB license record with its status and the date that status was read.
It is not a ranking and it is not a recommendation. Being named on an ADU permit is evidence of engagement in that jurisdiction on those dates.
It is not evidence of quality, of completion, or of work anywhere else. California licenses no ADU classification, so no row here says a company is an ADU specialist. Owner-builder permits are excluded from the counts.
adu-construction-duration-la.csv and adu-permit-times-by-city.csv — two different clocks
These measure different things and are routinely conflated. Permit times is application to issuance: how long the counter took. Construction duration is issuance to certificate of occupancy: how long the build took. Adding them does not give a project timeline, because a homeowner’s design and pre-application period sits before both.
The duration file is Kaplan-Meier: permits with no certificate of occupancy at the observation cut are right-censored rather than dropped, which is why it carries a median that is not the completed-only median. The completed-only figure is in the file too, labeled as a contrast. Use the censored one.
adu-declared-permit-valuations.csv — the most misused number in ADU coverage
Declared permit valuation is not construction cost. It is the dollar figure an applicant writes on a building-permit application, which cities use to compute plan-check and permit fees, and which some cities replace entirely with a published dollar-per-square-foot table.
Under-declaration is a known practice. A $20,000 garage-conversion valuation in Los Angeles is not a $20,000 project.
It is published because it is what the cities recorded and because readers are routinely shown it as "what ADUs cost". The distribution is the honest object; the label is the whole point. Owner-builder-named records are excluded and zero valuations are dropped, both stated in the file header.
adu-permits-vs-completions.csv — agency data, compiled not computed
HCD Housing Element Annual Progress Report figures for the twenty jurisdictions whose returns were usable, with permits, completions, the completion-per-permit ratio and how many years each reported. The provenance class is agency published: HCD produced these counts, we assembled them.
A permit is not a completion, and the gap between the two columns is the reason this file exists. Jurisdictions that reported permits but no completions in a year are visible in the year counts rather than hidden in an average.
adu-preapproved-plan-programmes.csv — coded from primary pages
35 named California jurisdictions, coded from each jurisdiction’s own published program page: whether a program exists, its status, how many plans, the plan fee, and what a pre-approved plan still does not remove.
A pre-approved plan is not a permit. Every row carries a "still requires" field for exactly that reason — site-specific zoning, foundation and utility review survive the pre-approval in every program we read.
adu-hud-fmr-fy2026.csv — agency data, four areas only
HUD Fair Market Rents for FY2026 in four California HUD Metro FMR Areas, by bedroom count, each row linking to its own HUD documentation page.
Fair Market Rent is a 40th-percentile gross rent policy construct including utilities. It is not an asking rent, it is not a market median, and it is not what a specific ADU will let for.
It is in the shelf because it is the defensible floor under a rental-income estimate, and because the alternative — an invented per-area rent — is what we exist not to publish.
adu-statute-change-log.csv and adu-statute-retired-numbers.csv
The section-level enactment register for California Government Code Chapter 13 — all 36 sections with the bill named on each section’s own enactment line and the effective date of the current text — plus the four retired section numbers with their successors and the renumbering authority quoted from the successor’s own line.
These two are the only files on the shelf that need no statistical caveat and every citation caveat. They are current to their retrieval date and to nothing else. The chapter has been amended by six bills in under three years.
The provenance label on every row
The same three classes the statistics hub prints on every figure and the methodology page defines in full. On this shelf the label sits on the file rather than the number, because a whole file has one origin.
- HyreADU calculation
- We downloaded a primary file — a municipal open-data extract, a license register, a statute — and computed the figures ourselves. The method is on the linked study page and the exclusions are in the file header. Seven of the nine files.
- Agency published
- HCD, HUD, the Census, CSLB or a city produced the number and we compiled it into a usable shape. Attribution belongs to the agency; the compilation is ours. Two of the nine files.
- Trade estimate
- A builder survey, a cost index or an owner survey. Useful, labeled, and never presented as a measurement of every ADU in California. No file on this shelf is one — trade estimates appear on study pages as cited comparisons, not as data we redistribute.
How to cite and reuse these files
Creative Commons Attribution 4.0. Commercial use is fine, redistribution is fine, derivative work is fine, and the only requirement is attribution.
- 1 Name the desk, the file and the retrieval date
The short form: HyreADU Research Desk, “adu-permit-times-by-city.csv”, retrieved [your date], hyreadu.com/data/adu-permit-times-by-city.csv, CC BY 4.0. The date matters more than it usually does, because several of these files are refreshed rather than versioned.
- 2 Carry the label, not just the number
If you reuse a valuation figure, carry the words “declared permit valuation” with it. If you reuse a permit count, carry “permits, not completions”. Stripping the label is how a fee base becomes a construction cost in somebody else’s headline, and we would rather be uncited than cited that way.
- 3 Attribute the agency where the agency produced it
For the two agency-published files, credit HUD or California HCD as the source and HyreADU as the compilation. We did not survey rents and we did not collect the annual progress reports.
- 4 Link the method page, not just the file
Every row of the inventory carries a “Method” link. A reader who has the file and not the frame will misread it, and the frame — the exclusions, the censoring, the date the feed stops — is on the study page.
- 5 Tell us if you find an error
Corrections are made in place, the review date changes, and where the substance changed we say what changed. If a figure you recomputed does not match ours, we want the discrepancy more than we want the quiet.
The shape of the files, and how to load them
All nine are UTF-8 CSV, comma-separated, quoted where a field contains a comma. Each file begins with a single comment line starting #, then one header row, then data.
The comment line is deliberate: a file that travels without its caveat is a file that will be misread, and a spreadsheet drops the caveat the moment it is separated from the page.
In pandas, pd.read_csv(path, comment='#') reads them correctly. In R, read.csv(path, comment.char = "#"). In a spreadsheet, the comment line arrives as a first row you can delete. Nothing in these files needs a parser flag beyond that.
Dates are ISO-8601 (YYYY-MM-DD) throughout, including in the statute register’s effective-date column. Money is whole US dollars with no symbol or separator. Rates and shares are decimals, not percentages — a completions-per-permit value of 0.5959 is 59.59 per cent.
Empty means unverified, never zero: a blank cell in these files is a cell we could not fill from a primary source, and it is left blank rather than filled from an aggregator.
What is deliberately not on this shelf
The absences are as much a part of the method as the files. Each of these is a dataset a reader might reasonably expect, and each is missing for a stated reason.
- A statewide city-by-city ADU fee table
The fee study met a retrieval wall: the platform hosting most California municipal codes returns HTTP 403 to automated requests, and several city fee schedules were unreachable. A fee file assembled from aggregator numbers would be the single most reused file on this shelf and the single least defensible. It is not published.
- A normalised ordinance table for every California city
Same wall, same reason. The ordinance tracker publishes the coding schema and the full state floor instead, plus the two jurisdictions whose own pages could be read. A table of every Californian jurisdiction filled from secondary sources is not a dataset, it is a liability.
- A national ADU dataset
HyreADU is a California program. The company store is 734 California firms and one out of state, and permit evidence exists for five named jurisdictions.
Nothing computed from that store can be titled to the United States, so no national file exists here.
National claims on this site rest on federal agency data and are cited to the agency.
- Anything containing a homeowner’s personal information
The permit feeds these files come from are public records and several carry applicant details. The published files carry contractor and jurisdiction fields only. No owner name, no site address, no parcel identifier is redistributed here.
- A cost-per-square-foot figure presented as a price
The valuation file contains declared valuations and the study computes rates from them, all labeled. There is no file on this shelf that answers "what does an ADU cost" with one number, because the honest answer to that question is a distribution with a caveat attached, and it is on the cost study.
Versioning, refreshes and what happens when a figure changes
The download files are refreshed, not versioned. A file at a given URL is the current computation, and when the underlying feed updates or an error is found the file at that URL changes.
There is no archive of superseded editions today, and pretending otherwise would be the easiest false claim on this page to make.
That has one consequence for anyone citing them, and it is the reason the citation format above puts the date first: record the date you downloaded the file. If your figure and ours diverge later, the date is what makes the divergence resolvable rather than an argument.
Corrections. Where a published figure turns out to be wrong, the file and the study page are corrected in place, the review date changes, and where the substance changed the study page says what changed and when. We do not silently amend a number that somebody may have cited.
Refresh cadence. The statute register is re-retrieved whenever a bill touching Chapter 13 is chaptered, which in the last two years has meant twice.
The municipal permit files depend on feeds that update on their own schedules, and at least one — the Los Angeles legacy contractor extract — stops in May 2023 by construction of the feed rather than by choice.
Every file states its own currency in its header comment and in the row above.
If you are a reporter working to a deadline
Take the file, take the method link next to it, and take the caveat sentence from the file header. Those three things together are everything you need to attribute a figure correctly, and the caveat sentence is written to be quotable as it stands.
If the cut you need is not in the published file — a different jurisdiction, a different year band, a different exclusion rule — say so rather than approximating it from what is here.
The desk will either compute it and publish it, or say that it cannot be computed from the sources we hold. The one thing we will not do is produce a number for a deadline that we would not publish on the site.
The press page sets out how the desk should be described, what it is and is not, and the route for a query. The methodology page is the full account of how these figures are made.
Questions
Are these files free to use commercially?
What format are the files in?
Does an empty cell mean zero?
Is the declared permit valuation file a construction cost dataset?
Why is there no ADU fee dataset?
Do these files cover states other than California?
Do the files contain personal information from permit records?
Are old versions of the files archived?
What if I recompute a figure and get a different answer?
Can you produce a cut that is not published here?
Written and audited by
HyreADU Research Desk
Primary-source research, data analysis and fact checking
We are a research desk, not a builder. We read the permit extract, the statute, the HCD return or the fee schedule ourselves, and publish each figure with its source and retrieval date.
Where a number cannot be traced to a primary source, we leave it out and say what we could not verify. Our store-based claims cover California only.
- CA
- the only state this desk will make store-based claims about
- 5
- jurisdictions with extracted ADU permit evidence
- 735
- CSLB-verified companies in the California store
- 0
- national claims from a one-state store
How this desk works
- Primary sources only. Permit counts come from the city or county that issued the permit. Production counts come from HCD’s Annual Progress Report. Rents come from HUD or the Census. We do not cite an article that cites a source; we download the source and compute the figure ourselves.
- This is a California site. The company store is 734 California firms and one New Mexico firm. Permit evidence exists for five named jurisdictions: Los Angeles, San Francisco, Sacramento, San José and unincorporated Marin. A number from that store is titled to those places, never to the United States.
- A permit is not a completion, and a license is not an ADU grade. California licenses no ADU classification. Being named on an ADU permit is evidence of engagement in that jurisdiction, not of quality, completion, or work anywhere else. Owner-builder permits are excluded from contractor counts.
- Calculation is labeled as calculation. Figures we derive are never presented as something HCD, HUD, the Census or a city published. Terner Center research is cited as Terner’s, never restated as ours.
- We do not design, permit or build ADUs, and we take no payment for placement, ranking or a favorable mention. Pages that look like rankings are not: they publish public-record counts and let the reader decide.
- Nothing here is legal, tax or financial advice. Zoning, underwriting and appraisal practice vary by jurisdiction, lender and appraiser. The useful next step on a specific lot is the planning counter and a licensed professional.
Data as of File inventory, row counts and byte sizes read from the published files 2026-09-07; each file’s own data currency is stated in its row and in its header comment. Authorship on this site is organizational: the analysis belongs to the desk rather than to a named individual, and we do not publish credentials we do not hold.
Our editorial policy sets out how we source, date and correct what we publish.
The data behind this page
Every figure on this page is computed from one file, and that file is published here so the analysis can be checked, disagreed with, or reused.
Sources & retrieval dates
- adu-statute-change-log.csv — California ADU statute change register , HyreADU calculation. 36 rows, 9 columns, 4.9 KB. Coverage: California, statewide. Period: 2024-03-25/2026-01-01. Underlying source: California Legislative Counsel, leginfo.legislature.ca.gov. Retrieved 2026-09-07. Method: /research/adu-ordinance-change-log/ Retrieved 2026-09-07.
- adu-statute-retired-numbers.csv — Retired California ADU statute section numbers , HyreADU calculation. 4 rows, 4 columns, 0.9 KB. Coverage: California, statewide. Period: 2024/2026. Underlying source: California Legislative Counsel, leginfo.legislature.ca.gov. Retrieved 2026-09-07. Method: /research/adu-ordinance-change-log/ Retrieved 2026-09-07.
- adu-permit-contractor-index.csv — ADU permit contractor index, five jurisdictions , HyreADU calculation. 3,668 rows, 14 columns, 445 KB. Coverage: Los Angeles, San Francisco, Sacramento, San José, unincorporated Marin. Period: 2016/2026. Underlying source: Five municipal open-data permit extracts, matched to CSLB license records. Retrieved 2026-08-30. Method: /research/adu-permit-contractor-index/ Retrieved 2026-08-30.
- adu-construction-duration-la.csv — ADU permit-to-occupancy duration, City of Los Angeles , HyreADU calculation. 24 rows, 10 columns, 1.8 KB. Coverage: City of Los Angeles. Period: 2020-01-01/2026-08-31. Underlying source: LADBS Socrata resource pi9x-tg5x. Retrieved 2026-09-05. Method: /research/adu-construction-duration/ Retrieved 2026-09-05.
- adu-permit-times-by-city.csv — Application-to-issuance permit times, three jurisdictions , HyreADU calculation. 17 rows, 9 columns, 3.1 KB. Coverage: City of Los Angeles, City and County of San Francisco, unincorporated Marin. Period: 2020/2026. Underlying source: Municipal open-data permit extracts. Retrieved 2026-09-05. Method: /research/adu-permit-times-by-city/ Retrieved 2026-09-05.
- adu-declared-permit-valuations.csv — Declared building-permit valuations, five jurisdictions , HyreADU calculation. 20 rows, 13 columns, 4.1 KB. Coverage: Los Angeles, San Francisco, Sacramento, San José, unincorporated Marin. Period: 2016/2026. Underlying source: Nested permit valuation fields in five municipal harvests. Retrieved 2026-09-05. Method: /research/adu-permit-valuations/ Retrieved 2026-09-05.
- adu-permits-vs-completions.csv — ADU permits against completions, top 20 usable jurisdictions , Agency published, compiled by HyreADU. 20 rows, 8 columns, 1.3 KB. Coverage: California, 20 jurisdictions. Period: 2018/2025. Underlying source: California HCD Housing Element Annual Progress Report. Retrieved 2026-09-05. Method: /research/california-adu-permits-vs-completions/ Retrieved 2026-09-05.
- adu-preapproved-plan-programmes.csv — Pre-approved ADU plan programs , HyreADU calculation. 35 rows, 8 columns, 14.9 KB. Coverage: 35 named California jurisdictions. Period: 2026. Underlying source: Each jurisdiction's own published program page. Retrieved 2026-09-05. Method: /research/pre-approved-adu-plans/ Retrieved 2026-09-05.
- adu-hud-fmr-fy2026.csv — HUD Fair Market Rents FY2026, four California FMR areas , Agency published. 4 rows, 9 columns, 1.4 KB. Coverage: Four California HUD Metro FMR Areas. Period: FY2026. Underlying source: HUD Fair Market Rent Documentation System. Retrieved 2026-09-05. Method: /research/adu-rental-income/ Retrieved 2026-09-05.
Take the file and the caveat together
Every file here carries a one-line header stating what it is and what it is not. That line is written to be quoted as it stands, and it is the difference between a figure that survives scrutiny and a figure that starts an argument.
HyreADU does not design, permit or build accessory dwelling units. These files are published for research and journalism and are informational: they are not legal, tax, financial, appraisal or construction advice.
The contractor index is not a ranking, a recommendation or a quality measure — being named on a permit is evidence of engagement in that jurisdiction on those dates and nothing more, and California licenses no ADU classification.
Declared permit valuation is not construction cost, a permit is not a completion, and Fair Market Rent is not an asking rent. Empty cells mean unverified, never zero.
Files are refreshed in place rather than versioned, so record the date you downloaded one. Every file is published under Creative Commons Attribution 4.0; the two agency-published files should additionally credit the originating agency.