ADUs as rental housing: what the rents actually look like

The question the category answers with a monthly figure

“What will it rent for?” is the second question a homeowner asks, after cost, and it is the question the category has answered with a round number — often a metro-wide one-bedroom, sometimes an unsourced “California ADU average,” occasionally HUD Fair Market Rent copied without the sentence that says what FMR is.

The number then becomes the numerator of a yield, the income line of a loan pitch, and the reason the project “pencils.”

We went looking for an observed ADU rent series: a current, statewide, bedroom-specific survey of what permitted accessory units actually let for. We did not find one that is still in the field.

What we found is HUD’s FMR, the Census Bureau’s ACS gross rent, and one 2021 owner survey whose occupancy finding is more important than its rent finding, because occupancy decides whether the rent is collected at all.

HyreADU analysis: FMR and ACS are real public series, and they are useful as area context for 0-, 1- and 2-bedroom stock that is the size of a typical ADU. They are not ADU rents.

The Terner/CCI occupancy split is the correction the yield pages skip. A digest of the figures in this study lives at ADU rental income statistics; this page is the method, the caveats and the worked example.

Current construction cost — the denominator — is the California cost study. How a lender will (and will not) count projected rent is the financing-landscape study. Whether the unit “adds value” is the appraisal study. Those are four pages because they are four questions.

What HUD Fair Market Rent actually is

The current published series, retrieved 5 September 2026 from HUD User, is fiscal year 2027.

HUD’s notice (Fair Market Rents for the Housing Choice Voucher Program, Moderate Rehabilitation Single Room Occupancy Program, and Other Programs, Fiscal Year 2027, 91 FR 56156, 1 September 2026) states that FY 2027 FMRs become effective on 1 October 2026 unless HUD receives a valid reevaluation request for a specific area.

Through 30 September 2026 the in-force series remains FY 2026, including the 21 April 2026 revision (91 FR 21301) that raised Los Angeles-Long Beach-Glendale, Napa and San Luis Obispo-Paso Robles on local survey data.

This study prints FY 2027, because that is the series HUD is now publishing and the series a reader after 1 October will look up.

We name both dates so a reader in the last weeks of FY 2026 is not misled.

HUD’s own definition, unchanged in substance across recent fiscal years and restated in the FY 2027 notice, is that “the FMR for an area is the amount that a tenant would need to pay the gross rent (shelter rent plus utilities) of privately owned, decent, and safe rental housing of a modest (non-luxury) nature with suitable amenities. HUD’s FMR calculations represent HUD’s best effort to estimate the 40th percentile gross rent paid by recent movers into standard quality units in each FMR area.” For FY 2027, HUD uses Census ACS 5-year data collected between 2020 and 2024 as the base rent, applies a recent-mover adjustment from ACS 1-year 2024 where it passes statistical tests, then trends forward with a gross-rent inflation factor.

Local PHA-conducted surveys replace the ACS base in named areas; the FY 2027 notice lists Los Angeles-Long Beach-Glendale, Napa and San Luis Obispo-Paso Robles among California areas with 2025 local survey data.

Bedroom columns are ratios applied to the two-bedroom FMR. A studio is HUD’s efficiency (0-bedroom) column.

That is why this page prints 0-, 1- and 2-bedroom FMRs and stops there: Chapple, Ganetsos and Lopez (2021) found 61 percent of new California ADUs were one-bedroom and 18 percent studios; units with two or more bedrooms were 21 percent. Three- and four-bedroom FMRs describe a different dwelling.

Small Area FMRs, published at ZIP-code level inside designated metropolitan areas, can sit far from the metro-wide number. Los Angeles-Long Beach-Glendale is a Small Area FMR area for Housing Choice Voucher programs.

A metro-wide 1-bedroom FMR applied to a backyard studio in a low-rent ZIP, or to a finished unit in a high-rent ZIP, is the naive lookup.

The honest lookup, if you are in a designated metro, is the Small Area column plus a look at listings on the block. We do not ship a ZIP table; HUD revises it, and a stale table would present itself as local knowledge.

The rent estimator is the page that refuses to invent your area’s figure.

ACS median gross rent, 2020–2024, named California markets

Census table B25064 is median gross rent in dollars, universe renter-occupied housing units paying cash rent. The current 5-year vintage is 2020–2024, released 29 January 2026.

Census QuickFacts series HSG860224 is the Bureau’s public digest of that table for states, counties and places. Figures below are that digest, retrieved 5 September 2026. They are all rental housing, not ADUs.

Households not paying cash rent are excluded from the median, which matters for this study: a relative living rent-free in an ADU is exactly the household B25064 drops.

California’s 2020–2024 median gross rent is $2,036. The United States median is $1,413. California is not a national rental market, and a statewide median is not a 1-bedroom number: it mixes studios in the Central Valley with three-bedrooms on the coast. The useful rows are the named counties.

Among the counties that overlap the FMR areas above, 2020–2024 median gross rent runs from $1,392 in Fresno County to $2,922 in San Mateo County and $2,857 in Santa Clara County.

Los Angeles County, the largest rental market in the state and a single-county HUD Metro FMR Area, sits at $1,954 — below the statewide median, because the statewide median is pulled up by the Bay Area. The City of Los Angeles is $1,933.

San Francisco County is $2,476. San Diego County is $2,246. Orange County is $2,434. Sacramento County is $1,790. Riverside County is $1,901; San Bernardino County is $1,801. Alameda County is $2,357; Contra Costa County is $2,375.

Those are ACS medians of all renter-occupied units paying cash rent. They are not 1-bedroom rents.

Santa Clara County’s $2,857 median sitting above the FY 2027 San Jose-Sunnyvale-Santa Clara 1-bedroom FMR ($2,778) is what a mixed-bedroom median does in a market where two- and three-bedroom units are common and expensive.

Los Angeles County’s $1,954 median sitting below the FY 2027 Los Angeles-Long Beach-Glendale 1-bedroom FMR ($2,402) is what a 40th-percentile recent-mover FMR, trended, does relative to a five-year in-place median. Different universes. Different percentiles.

Different years inside the window. The comparison is the finding; a blended “California rent” is not.

Tenure and structure: the host-lot stock, labeled as all housing

ACS table B25003 is tenure of occupied housing units: owner-occupied versus renter-occupied. Census QuickFacts “owner-occupied housing unit rate, 2020–2024” (HSG445224) is that table as a percentage of occupied units.

California’s rate is 55.9 percent owner-occupied (so 44.1 percent renter-occupied). It is not uniform. San Francisco County is 38.2 percent owner-occupied. The City of Los Angeles is 36.0 percent. Los Angeles County is 45.9 percent.

Santa Clara County is 55.1 percent, Alameda County 54.4 percent, San Diego County 54.6 percent, Fresno County 55.5 percent. An ADU is typically added to an owner-occupied lot.

The renter share in the surrounding stock is the market the unit would join if it were let — not a vacancy rate, and not a forecast of who will live in the accessory unit.

ACS table B25024 is units in structure. Accessory dwelling units are not a Census category.

A detached backyard cottage may be reported as a 1-unit detached housing unit; an attached ADU or JADU may be folded into the primary dwelling; a garage conversion may be either.

The table cannot be used as an ADU count, and we do not.

It is here as the shape of the housing stock that FMR and B25064 are taken from, so nobody mistakes a 50-or-more-unit elevator building’s rent for a backyard cottage.

The detailed B25024 cells we retrieved from data.census.gov are ACS 2019–2023 5-year (table ACSDT5Y2023.B25024): California had 14,532,683 housing units, of which 8,315,954 (57.2 percent) were 1-unit detached and 1,061,574 (7.3 percent) were 1-unit attached.

The United States 1-unit-detached share in the same vintage is 61.4 percent (87,350,644 of 142,332,876).

The 2020–2024 5-year detailed B25024 table was not retrieved as a cell dump on this date; Census Reporter’s 2020–2024 5-year profile comparison puts California total housing units at 14,644,735, a modest increase on the 2019–2023 total.

Structure mix at the state level does not jump a year. The vintage of the cell counts is 2019–2023, and it is labeled.

HyreADU analysis: 57 percent 1-unit detached is the lot type state ADU law was written for. It is not 57 percent ADUs. FMR and ACS gross rent in those neighborhoods are still apartment-and-house rents.

The ADU-specific occupancy evidence is the survey in the next section, and it is a survey of people who built, not of the ACS housing stock.

Who lives there: the Terner/CCI owner survey

Karen Chapple, Dori Ganetsos and Emmanuel Lopez, Implementing the Backyard Revolution: Perspectives of California’s ADU Owners (UC Berkeley Center for Community Innovation, 22 April 2021), is the paper.

The PDF is at aducalifornia.org; the Terner Center published a same-day commentary, “First Ever Statewide ADU Owner Survey Shows Growth, Room for Improvement,” linking that PDF. David Garcia and Ben Metcalf of the Terner Center reviewed the report. It is CCI’s survey. It is cited here as Terner/CCI, never as a HyreADU finding.

Method, from the paper: a digital survey, English and Spanish, in late summer to fall 2020.

Recipients were identified from HCD Annual Progress Report addresses for ADU permits or certificates of occupancy in 2018 or 2019, supplemented with completed-ADU lists from named Bay Area jurisdictions.

Postcards went to 15,745 households; 823 responses; after dropping non-owners, a 4.8 percent response rate. The analytical sample of owners is 752.

Responses by region: San Francisco Bay Area 391 (52.0 percent), Los Angeles County 178 (23.7 percent), Central Coast 88 (11.7 percent), Orange and San Diego Counties 57 (7.6 percent), other 38 (5.0 percent).

The authors state the limitation themselves: the sample over-represents the Bay Area and under-represents Los Angeles County, and it covers only 2018–2019 completions, before the 1 January 2020 state-law round.

Rental tenure, executive summary, quoted: “About half (51%) of California’s new ADUs serve as income-generating rental units, and 16% of ADUs provide no-cost housing to a relative of the homeowner.” “Only 8% of new ADUs in California are short-term rentals, though more affluent homeowners are more likely to list their ADUs as short-term rentals than those making less than $100,000 a year.” The body (p. 14) adds that ADUs “often serve as affordable housing for friends or relatives: 18% of the state’s new ADUs provide no-cost housing for family members (16%) or friends (2%),” and that “approximately 16% of the new ADUs serve as home offices, studios, or guest houses for the property owner (Figure 9).” Figure 9 itself is asked only of the 49 percent who were not then renting out the unit (n = 366): 34 percent said a friend or relative is staying there for free; 34 percent used the space as something other than an apartment; 15 percent said it needs physical work to be rentable; 13 percent live in the ADU and rent the main residence; 2 percent vacant and looking; 3 percent other.

One internal tension is recorded, not smoothed. The executive summary’s “8% of new ADUs are short-term rentals” and the body’s “of the ADUs that are rented out to tenants, only 8% function as short-term rentals” are different denominators.

We quote both. We do not pick one and silently convert it. The Terner Center commentary uses the executive-summary form: “Only 8 percent of new ADUs in California are short-term rentals.”

Rents they reported, among those who charged rent, are dated and are not a 2026 market. Statewide median $2,000 a month; $3.68 per square foot.

By region (Figure 10, n = 225): Central Coast $1,925, Los Angeles County $2,000, Orange and San Diego Counties $2,150, San Francisco Bay Area $2,200. By bedrooms (Table 4): studio $1,800, 1-bedroom $2,000, 2-bedrooms $2,800, 3 or more $2,800.

By type: detached $2,200, addition $2,000, garage $1,875, conversion $2,200. Physical stock in the sample: 53 percent detached, 23 percent garage or other building conversion, 13 percent attached addition, 9 percent basement/attic/room conversion; average 615 square feet; 61 percent one-bedroom, 18 percent studio, 21 percent two or more bedrooms.

Median construction cost $150,000, or $250 per square foot — a 2021 report of 2018–19 invoices, not current cost. Current cost is the cost study.

HyreADU analysis: 51 percent generating rental income is not a vacancy rate for a landlord who intends to let. It is the share of a 2018–19 owner cohort who were collecting rent when surveyed in 2020.

The complementary share is family, friends, owner use, unfinished space, and a small vacant-and-looking sliver.

A yield model that puts market rent on 100 percent of ADUs is describing a different stock than the one the only statewide owner survey measured. That is the finding.

It is Terner/CCI’s measurement, dated, with a Bay Area-heavy sample, and it is still the occupancy evidence the category has.

A worked yield at three occupancy assumptions — not a forecast

The exercise is arithmetic on a named FMR, under three occupancy assumptions, on a dated construction-cost denominator. It is not a recommended return, not a cap rate, and not financial advice.

Vacancy, management, insurance, the Proposition 13 new-construction tax increment, maintenance and debt service are all off.

The ROI calculator is the page that runs tight, base and loose on a cost you type; this section shows why the occupancy toggle moves the numerator more than a bedroom-column choice.

The unit. A 1-bedroom ADU, because that is 61 percent of the CCI 2021 sample. The FMR is HUD FY 2027, 1-bedroom, Los Angeles-Long Beach-Glendale HMFA: $2,402 a month.

Annualised gross, 12 × $2,402 = $28,824. We also print San Francisco HMFA ($3,003), Oakland-Fremont ($2,290), San Jose-Sunnyvale-Santa Clara ($2,778), San Diego-Chula Vista-Carlsbad ($2,335), Sacramento--Roseville--Arden-Arcade ($1,793) and Fresno ($1,363) in the table, same assumptions.

Efficiency and 2-bedroom columns are in the FMR table above; we do not repeat the whole grid as a yield.

The denominator, labeled as dated. Chapple, Ganetsos and Lopez (2021) reported a statewide median construction cost of $150,000. That is a 2020 survey of 2018–19 invoices.

Using it under a 2027 FMR overstates gross yield, because costs have moved and the cost study is the current range.

We use $150,000 anyway, as the only statewide owner-reported median with a method, so the occupancy toggle is visible on a published denominator rather than on a cost we would invent. Current cost is the cost study.

Paste a cost you believe into the ROI calculator if you want a 2026 denominator.

Assumption A — market FMR, every month. The unit is let at the FY 2027 1-bedroom FMR, 12 months, no vacancy. Los Angeles: $28,824 a year.

Gross yield on $150,000: 19.2 percent. This is the brochure. It is also the assumption that is wrong for a large fraction of owners.

It is gross of vacancy; Fannie Mae’s purchase-file vacancy haircut on ADU rent used to qualify is 25 percent, which is an underwriting rule, not an occupancy study — see the financing-landscape study.

Assumption B — Terner haircut. Apply the CCI 2021 income-generating-rental share (51 percent) to the same FMR, as a stock expected-value: if you drew a random ADU from that cohort, 51 percent were collecting rental income.

Los Angeles expected annual rent: 0.51 × $28,824 = $14,700. Gross yield on $150,000: 9.8 percent. This is not a vacancy rate for an owner who will in fact let at market.

It is the correction to the claim that “ADUs” as a class throw off market rent. Short-term rental is not converted to FMR in this column; we will not invent a nightly rate.

Assumption C — rent-free family use. Sixteen percent of the CCI cohort housed a relative at no cost. Annual rent: $0. Gross yield: 0 percent.

The unit can still be housing, and it can still have a construction cost, an assessed increment and a contributory market value. It has no NOI.

Capitalising a market rent the owner will not collect, then calling the result “what the ADU is worth,” is the category error the appraisal study already refuses.

What a homeowner can actually do with this

Look up the FY 2027 FMR (and Small Area FMR if you are in a designated metro) at HUD User, then type it into the rent estimator with a vacancy percent you believe and any utilities you will pay. The tool will not invent the FMR.

After 1 October 2026, FY 2027 is the in-force series; until then FY 2026 remains in force and the two columns in our FMR table show how much that distinction is worth in Los Angeles versus Oakland.

Do not treat that result as a lease. Check listings for legal accessory units on your block, at your finish level. If you cannot find three, you have learned the same thing Freddie Mac’s ADU rental-comparable rule already admits: the comps are scarce.

Decide, in writing, whether the unit is for a tenant at market, for a tenant at below-market, or for a person who will pay nothing. Run all three in the ROI calculator on the same cost.

If the project only “works” in column A, it does not work for the fraction of owners the CCI survey actually measured, and it may not work for you once vacancy, tax and a 2026 invoice are in the sheet.

Budget the tax increment and the financing constraint separately. California will reassess the newly constructed portion; that is not a rent.

Fannie Mae and Freddie Mac will not count rent from a unit that does not yet exist, and they will cap the rent they do count. Those pages are linked above.

This page will not originate a loan and will not appraise the unit.

HyreADU analysis: the category has been answering an occupancy question with a market rent because occupancy does not close a lead. Occupancy is still the evidence.

We would rather publish the shorter honest page — FMR as FMR, ACS as ACS, Terner as Terner, three occupancy assumptions on a dated denominator — than an average California ADU rent we cannot source.