Grounded Research Note 1, August 2026 – Person-level measure of Australian home ownership, 2006 to 2021.Â
Summary
Australia’s published home ownership rate is a dwelling measure. It divides owner-occupied dwellings by all dwellings and contains no people.
Counted by person, using the same Census, the share of Australian adults who own the home they live in was 51 per cent in 2021, against the 66 per cent published.
The measure has been built on an identical construction at four censuses. The gap between the two has been 16 to 18 percentage points at every one, for fifteen years.
Across those four censuses the published rate has run about a third higher than the person-counted rate.
Close to three million Australian adults live in a home that somebody else in their household owns. They are recorded as neither owners nor renters, and they raise the published rate rather than lowering it.
No new Census question is required. Every variable has been collected since 2006. What has never been published is the two together.
What the published rate measures
The Census asks each household a single question about tenure and records a single answer. If one resident owns the dwelling, the dwelling is classified as owner-occupied. Every other adult under that roof is then carried on the owner side of any rate built from that classification.
This is a correct answer to the question the Census asks. The difficulty arises when the resulting rate is reported as though it described people, which is how it is almost universally used in public debate and in policy evaluation.
Definition
An adult is counted as a home owner if they are the household reference person, or that person’s resident spouse or partner, in a dwelling that is owned outright, owned with a mortgage, or being purchased under a shared equity or rent-buy scheme. The rate is that count divided by the population aged 18 and over.
The definition follows the homeowners-to-population ratio published by the Federal Reserve Bank of Minneapolis in July 2026, which moved the United States rate from 65 per cent to 53.
| Census | Adult owners | Adults 18+ | Counting people | Counting houses | Gap |
|---|---|---|---|---|---|
| 2006 | 8,203,711 | 15,262,825 | 53.7 | 70.2 | 16.4 |
| 2011 | 8,721,495 | 16,708,362 | 52.2 | 68.7 | 16.5 |
| 2016 | 9,072,157 | 18,471,296 | 49.1 | 67.2 | 18.1 |
| 2021 | 10,179,329 | 19,955,210 | 51.0 | 66.9 | 15.9 |
Per cent. Both series built from ABS Census TableBuilder Pro on an identical construction at every census. The 2021 person figure is lifted roughly a point by the closed international border, which removed several hundred thousand mostly non-owning temporary residents from the count.
The 2021 reading is inflated by the closed international border, which removed several hundred thousand temporary residents from the count. They are almost all adults and almost none of them owners. Returning half a million of them to the denominator puts 2021 below 50 per cent. The 2026 Census was taken with the borders open and the effect should reverse.
Validation
For 2021, household reference persons in owner tenures total 6,114,783. Owner-occupied dwellings, tabulated separately from tenure by number of persons usually resident, total 6,114,795. Since every owner-occupied dwelling contains exactly one reference person the two counts should be identical, and they are twelve records apart on a base of six million. That residual is consistent with the ABS confidentiality adjustment alone.
Each cell in TableBuilder is perturbed independently, so the same quantity recovered from different tables differs by a handful of records. Reading owner reference persons from the tenure column totals rather than from the reference person rows gives 6,114,776, a further seven records away. Nothing in the results depends on which is used.
The measured couple share of owner households moves smoothly across the series: 68.6 per cent in 2006, 67.8 in 2011, 67.4 in 2016 and 66.5 in 2021. In earlier estimates this quantity was the largest single assumption. It is now counted.
Who the non-owning adults are
The residual is not a renter population. About 28 per cent of adults rent. Roughly another 15 per cent live in a home that somebody else in their own household owns: adult children who have not been able to leave, flatmates, a parent who has moved in with their children, someone who separated and returned to a spare room, a person with disability living with ageing parents. The remainder are in aged care, other institutions, other tenures, or have no usual address.
The middle group is the entire difference between the two measures. They are recorded as neither owners nor renters, and the dwelling-based rate cannot see them.
Limits
The Census does not record who holds legal title. Both members of a couple are counted, following the Minneapolis convention. A title-strict measure would be roughly two points lower, so this assumption inflates ownership and understates the gap.
Property owned away from the usual residence is out of scope. The measure asks whether a person owns the home they live in, not whether they own property.
Cells carry small random adjustments for confidentiality. Reproduction will differ by a handful of records and not more.
Place of enumeration counts people where they were on Census night rather than where they usually live.
The shared equity and rent-buy tenure categories are named differently across cycles and are small in every year.
International context
The convention is not Australian. The United States, Canada and the United Kingdom publish the same dwelling fraction. Eurostat uses a population denominator but assigns the dwelling’s tenure to everyone inside it, which produces the same effect by a different route.
New Zealand is the exception. Stats NZ asks individuals directly whether they own or part-own the home they live in, and has published individual home ownership for years. Its 2023 Census was the last traditional one; from 2030 New Zealand moves to administrative data.
The request
That the ABS publish, as standard Census output alongside the dwelling-based rate, the share of Australian adults who own the home they live in. No new question, no change to the form, no additional burden on respondents.
On 14 August 2026 the ABS advised Grounded that it recognised the importance of the topic and was considering it as part of the planning for next year’s data release.
Update, 26 August 2026. The ABS has confirmed that every variable this measure needs is present and unchanged in the 2026 Census, and that home ownership is on its list of possible analytical articles for the June 2027 release. Nothing new has to be collected. The published rate counts dwellings and is read as though it counted people. We are asking the ABS to publish the person-level figure alongside it, so the difference is visible at the point the number is used.
How this figure changed
When Grounded first published this work, on 10 August 2026, the person-counted rate was given as about 52 per cent. That was an estimate built from published ABS aggregates, carried with a stated range of 49 to 55, and it rested on one load-bearing assumption: the share of owner-occupied households containing a couple, which the Census records but the ABS does not publish.
TableBuilder Pro access was granted on 9 August and the measure was built directly rather than estimated. The couple share is 66.5 per cent. The rate is 51. The estimate had been right to within a point, and the assumption it depended on no longer exists.
The original media release remains online unchanged, with a dated update note recording the correction.
Appendix: cell counts as extracted
Every figure in this note can be rebuilt by anyone with a TableBuilder account, which is free. The counts below are as extracted on 12 August 2026 and are sufficient to reproduce every result above.
| Quantity | 2006 | 2011 | 2016 | 2021 |
|---|---|---|---|---|
| Reference persons, owned outright | 2,430,731 | 2,488,147 | 2,565,692 | 2,872,330 |
| Reference persons, owned with a mortgage | 2,419,724 | 2,692,366 | 2,844,711 | 3,228,376 |
| Reference persons, shared equity or rent-buy | 16,390 | 17,065 | 10,513 | 14,070 |
| Reference persons in owner tenures | 4,866,845 | 5,197,578 | 5,420,916 | 6,114,776 |
| of which, with a resident partner | 3,336,866 | 3,523,917 | 3,651,236 | 4,064,546 |
| All reference persons, being occupied private dwellings | 7,144,098 | 7,760,323 | 8,286,076 | 9,275,219 |
| Tenure not stated | 206,719 | 195,211 | 224,868 | 136,540 |
| Persons aged 18 and over | 15,262,825 | 16,708,362 | 18,471,296 | 19,955,210 |
| Counting people, per cent | 53.7 | 52.2 | 49.1 | 51.0 |
| Counting houses, per cent | 70.2 | 68.7 | 67.2 | 66.9 |
Owners are twice the partnered reference persons plus the remainder, divided by persons aged 18 and over. The dwelling-based rate is reference persons in owner tenures divided by all reference persons less tenure not stated. Each cell is perturbed independently for confidentiality, so column totals and row sums differ by a handful of records.
The category selections are set out in the method table above. Grounded will supply the saved TableBuilder outputs to anyone who asks.
| Element | Selection |
|---|---|
| Product | ABS Census TableBuilder Pro, accessed 12 August 2026 |
| Datasets | 2006, 2011, 2016 and 2021 Census, counting persons, place of enumeration |
| Tenure | TEND: owned outright; owned with a mortgage; and the shared equity or rent-buy category as named in each cycle |
| Reference person | RPIP: reference person in primary family, and reference person in non-family household |
| Partner present | 2021: SPIP. 2006 to 2016: RLHP husband, wife or partner categories, which the earlier cycles carry in place of SPIP |
| Denominator | AGEP, persons aged 18 and over, same dataset and vintage |
| Arithmetic | Owners = 2 × (partnered reference persons in owner tenures) + (non-partnered reference persons in owner tenures) |
Second and third family reference persons are excluded, since only one person per dwelling can be the household reference person. The measure can only be built at place of enumeration, because the person-level file on place of usual residence carries no tenure variable.