What can be known with high confidence from this data?
Observed ACS base: the strongest fact is the owner-occupied fee-payer base. Across the full national PUMA universe, 21.67M owner-occupied units report a monthly condo or HOA fee, and an estimated 53.79M residents live in those observed owner fee-paying units.
Observed fee level: the weighted average reported monthly assessment for those owner fee-paying units is about $258. Housing units, tenure, property type, PUMA/state geography, rent, home value, and household-reported fee amounts are also survey-derived measures.
Not directly observed: renter association units, vacant association units, exact legal association counts, HOA versus condo versus co-op legal type, association names, addresses, CAM providers, management-company use, and the CAM-company fee share are not directly available in ACS/IPUMS.
How does this app estimate total association units?
Step one: the model starts with owner-occupied units where ACS reports CONDOFEE greater than zero. That is the directly observed fee-paying owner base.
Step two: renter and vacant units are multiplied by separate AHS-calibrated rates for the same PUMA and property type. Direct AHS metro evidence is used where available; Census-division or national evidence supplies explicit fallbacks.
Current result: the national universe contains 21.67M observed owner fee-paying units, 6.56M estimated renter association units, and 2.99M estimated vacant association units, for 31.22M estimated association units.
Primary risk: AHS is a weighted survey rather than an association registry, and fallback geographies borrow broader-market rates where a direct metro estimate is unavailable.
Should the app align to FCAR's 2025 estimates?
Use FCAR as a benchmark, not a forced answer: FCAR is useful because Board and CEO audiences may know it, but its figures are curated estimates from multiple data streams, not a transparent reproduction of this app's ACS/IPUMS model.
Denominators differ: FCAR's 29.6M units divided by 35.2% implies a denominator of about 84.1M units, not all U.S. housing units. The app's 21.3% equals 31.22M estimated association units divided by 146.74M housing units across the full national PUMA universe. The percentages are therefore not directly comparable without first aligning the denominator and association definition.
Methods differ: this app directly observes owner fee-paying units and estimates renter and vacant exposure from AHS-calibrated rates. It cannot separately identify every cooperative or legal community-association form, while FCAR is a curated national estimate assembled through a different methodology.
Best presentation: show both layers: an observed ACS owner fee-payer base and a modeled association-market estimate. A calibrated FCAR benchmark can be shown as an external comparison, but it should not replace the transparent ACS calculation.
How should Annual Market Revenue be interpreted?
Revenue opportunity: managed association units x applied average monthly assessment x 12 x CAM fee assumption. It is not total assessments collected by associations, and the CAM fee and self-managed inputs are scenarios rather than observed contract terms.
How were assessment fees obtained?
Observed household field: assessment fees come from the Census ACS housing PUMS field CONP, the source equivalent of IPUMS USA's CONDOFEE variable. In 2024, it reports the monthly condominium and/or homeowners association fee for owner-occupied homes.
Weighted aggregation: the app keeps owner-occupied records where CONDOFEE is greater than zero, counts those as observed owner fee-paying units, and weights each reported monthly fee by the ACS household weight. The average monthly assessment is the weighted mean of those owner-reported fees; the median and percentiles are weighted quantiles of the same records.
Not CAM revenue: this is the household-reported monthly assessment paid to the association. It is not the management-company fee, not the association's full budget, and not separately observed for renter or vacant units.
How reliable are small property segments?
Confidence is sample-size based: High is 100+ unweighted owner fee-paying ACS records, Medium is 30-99, and Low is fewer than 30.
Scope is limited: the confidence label grades the owner assessment sample only. It does not measure uncertainty in renter/vacant assumptions, PUMA-to-metro allocation, or the CAM fee and self-managed scenarios.
Why do property filters change residents and economic fields?
Filters change the denominator: resident totals sum ACS residents whose housing records match the selected property segment; home value and rent use records from that same segment.
Does Residents mean people living in association units?
No: the Market Context Residents figure counts all residents in the selected geography and property segments. The app can directly count 53.79M residents in observed owner fee-paying homes nationally, but it does not currently estimate residents in modeled renter or vacant association units.
Is Observed owner home value a directly measured portfolio value?
No: it is a proxy equal to observed owner fee-paying units multiplied by the average value of all owner-occupied homes in the same geography/property segment. ACS identifies the fee-paying units, but the displayed valuation is not restricted to those homes and should not be compared directly with a registry-based portfolio value.
Can this identify specific associations or CAM providers?
No entity lookup exists: public ACS microdata does not show association names, addresses, manager relationships, portfolios, or CAM vendors.
What does Boat, RV, van, etc. mean?
Official Census category: "Boat, RV, van, etc." is the U.S. Census Bureau label for ACS BLD=10. The 2024 source contains 199,417 such units but does not identify boats, recreational vehicles, vans, or similar occupied housing forms separately.
Why can mobile or manufactured homes show modeled association units but no observed owner fee units?
The source fields differ: the 2024 housing PUMS contains no positive owner assessment observations in the mobile/manufactured structure category, so its directly observed owner fee-paying count is zero. The app nevertheless estimates about 398,120 renter and vacant association units from AHS-calibrated rates.
Interpretation: those units and any related revenue are fully modeled, and their assessment uses the documented geographic fallback chain. They should be treated as a directional scenario rather than an observed mobile-home association market.
Why does revenue not equal managed units times the displayed average assessment?
Weights differ: revenue uses an applied assessment for every modeled association segment, including explicit fallbacks where that segment has no owner fee sample. The general average assessment is weighted only by observed owner fee-paying units.
Default national reconciliation: 31.22M association units x 65.0% managed = 20.29M managed units. The applied monthly assessment is about $289.26, so 20.29M x $289.26 x 12 x 7.5% = about $5.28B.
Which assumption matters most to revenue?
CAM fee sensitivity: revenue changes in direct proportion to the CAM fee assumption. A 1% relative increase raises revenue 1%; a one-percentage-point increase from 7.5% to 8.5% raises revenue 13.3%.
Self-managed sensitivity: revenue changes with the managed share, equal to 100% minus the self-managed assumption. Raising self-managed units from 35% to 36% reduces the managed share from 65% to 64%, lowering revenue by about 1.54%.
How should cross-state metros be reviewed?
Grouped metros show total market size: ungrouped state slices are better for state-level regulation, operating coverage, or sales-territory planning.
Why can property-type assessments differ from metro averages?
Property-type fees use segment records: detached, townhouse, small multifamily, mid-rise, and large multifamily homes have separate observed owner-fee samples.
Why do property types have different revenue economics?
Three inputs vary: association-unit penetration, monthly assessment level, and managed-unit exposure differ materially across structure categories.
What should not be concluded from this tool?
Do not infer precision beyond ACS: exact association counts, manager identity, addressable portfolios, and sub-metro sales territories are not observable here.
Is this using the latest available relevant data?
Yes, for this model: the app uses 2024 ACS 1-year Census housing PUMS and a matching IPUMS person extract. IPUMS added the 2024 1-year ACS/PRCS sample on December 18, 2025, and that release is the first 1-year ACS/IPUMS sample where CONDOFEE reports homeowners' association fees in addition to condominium fees.
Why not the 2024 5-year file: IPUMS later added the 2024 ACS 5-year file, but it blends 2020-2024 records. The 2020-2023 records use the older, narrower condominium-fee concept, so the 5-year file would dilute the HOA signal this app needs.
Why not 2025: 2025 ACS 1-year should also be relevant once available because the expanded CONDOFEE definition should continue, but IPUMS has not shown 2025 ACS 1-year as available in the revision history checked for this app.
