Australia's oldest suburbs have the same size houses as its youngest. They just have half a person fewer in them.

There are 21 suburbs in Australia where the houses are emptier than anywhere else in the country. Three-bedroom homes with two people in them, street after street. They are spread across five states and most of them are on the coast.

The interesting part is not that those homes are large. They are not. The houses in Australia's oldest suburbs are the same size as the houses in its youngest — slightly smaller, in fact. What has changed is how many people are inside them.

Treasury published its Intergenerational Report on 21 September 2026, with a chapter on ageing and a separate chapter on housing. Both are national projections about 2066. We hold the 2021 and 2016 Censuses for every suburb in Australia, so we can ask a smaller question with a firmer answer: where has the ageing already happened, and what did it do to the houses?

The 21 emptiest suburbs in Australia

The ABS publishes a figure for each suburb called average persons per bedroom. Nationally it sits at 0.83. Of the 3,025 suburbs with at least 2,000 residents, exactly 21 come in at 0.6 — the lowest value anywhere in the country at that size.

Suburb	State	Over 65	Population
Tea Gardens	NSW	62%	3,288
Encounter Bay	SA	53%	5,305
Harrington	NSW	49%	3,381
Sorrento	VIC	49%	2,013
McCracken	SA	49%	2,076
Erskine	WA	45%	5,429
Tuross Head	NSW	45%	2,353
Portarlington	VIC	44%	4,436
Point Lonsdale	VIC	43%	3,788
Burradoo	NSW	42%	2,879
Twin Waters	QLD	42%	2,966
Bundanoon	NSW	41%	2,869
Moonta Bay	SA	40%	2,633
Banksia Beach	QLD	39%	7,180
Tura Beach	NSW	39%	3,405
Blairgowrie	VIC	39%	2,786
Mollymook Beach	NSW	37%	2,531
Dudley Park	WA	33%	6,957
St Leonards	VIC	33%	3,542
Vincentia	NSW	29%	3,870
Geographe	WA	27%	3,622

They hold 77,309 people between them across five states. Their average over-65 share is 41.9%, against 17.2% for the country. Two in five residents are past retirement age, and the list reads like a map of the Australian coast — Tea Gardens, Encounter Bay, Point Lonsdale, Mollymook Beach, Moonta Bay.

These are not the cheapest suburbs in Australia and several are not cheap at all. Burradoo and Sorrento are expensive. What they share is not price. It is that the houses have more bedrooms than the households have people.

The gradient, not just the extreme

Twenty-one suburbs could be a quirk. The pattern holds across all 3,025.

Sort every suburb by its over-65 share and split into five equal groups. The correlation between over-65 share and persons per bedroom is −0.50 — as a suburb gets older, its homes get emptier, steadily and without exception at the group level.

Over-65 quintile	Mean over 65	Bedrooms per home	People per home	Difference
Q1 (youngest)	8.2%	3.25	2.79	0.46
Q2	13.7%	3.18	2.66	0.52
Q3	17.0%	3.21	2.60	0.61
Q4	20.6%	3.23	2.52	0.71
Q5 (oldest)	29.0%	3.13	2.29	0.84

Read the bedroom column first, because it is the surprising one. It barely moves. The oldest fifth of Australian suburbs has 3.13 bedrooms per home; the youngest fifth has 3.25. If anything the oldest suburbs have slightly smaller houses.

The people column is where everything happens: 2.79 down to 2.29. Half a person fewer per home, across the same housing stock.

This matters because it removes the easy explanation. The obvious reason older suburbs might look under-occupied is that older suburbs have grand old houses. They do not. They have ordinary houses with fewer people in them.

So where did the people go? Nowhere — they were never there

The obvious explanation is that these suburbs aged in place: families who raised children in those houses stayed on after the children left, and the bedrooms emptied around them. That is what we assumed, and the first draft of this article argued it.

We tested it, and it is wrong. What follows is the test, and it is the reason the article says the opposite of what we set out to write.

The Census asks everybody where they lived five years earlier. Nationally, 56.6% of people were at the same address in 2021 as in 2016. If the emptiest suburbs were full of long-settled households, that number should be much higher there.

It is lower.

	Suburbs	At the same address 5 years earlier
The 21 emptiest suburbs	21	55.3%
Suburbs of the same age, not the emptiest	299	58.1%
All suburbs of 2,000+ people	3,024	56.5%

People in Australia's emptiest suburbs are slightly less rooted than the country as a whole, and clearly less rooted than other suburbs with the same share of residents over 65. Whatever emptied those houses, it was not a population sitting still.

On its own a gap of under three points across twenty-one suburbs would be thin evidence. The population figures settle it. Between the two Censuses, all 21 grew — median growth of 11.5% against 7.2% for the country — and they aged while doing it, a median shift of +3.9 points in over-65 share against +1.6 nationally.

That combination is the whole answer. A suburb ageing in place ages while its population sits flat or falls, because the ageing is the same people getting older. These suburbs got older and substantially bigger at the same time, which only happens when the people arriving are older than the people already there. Sixteen of the twenty-one grew faster than Australia did.

The pattern across the whole country says the same thing. Sort suburbs by age and the stayer rate rises steeply at first — 45.6% in the youngest fifth, where renters and young families move constantly — peaks at 60.7% in the second- oldest fifth, and then falls back to 59.1% in the oldest fifth. The very oldest suburbs are not the most settled ones.

They are places people retire to, two at a time

Read the list again and it is a list of destinations: Tea Gardens, Encounter Bay, Point Lonsdale, Mollymook Beach, Sorrento, Moonta Bay. These are the places Australians move to when they stop working, and the figures say the residents did exactly that. Tea Gardens grew 14% and its over-65 share rose more than ten points in five years, to 62%. Two of the twenty-one are exceptions worth naming: Sorrento and St Leonards in Victoria grew 26% and 43% and got marginally younger, which is what fast growth looks like when some of the arrivals are not retirees.

One more number finishes the picture. Of Australians aged over 65, 22.6% live alone — and that barely changes with how old a suburb is, moving only from 20.5% in the youngest fifth of suburbs to 24.5% in the oldest. Older people living alone are not concentrated in retirement areas at all; they are disproportionately in inner-city apartments. The retirement coast is couple territory.

Put the three findings together and the arithmetic resolves. A couple in their sixties or seventies moves to the coast and buys a three-bedroom house, because that is what is built there and because the grandchildren visit. Two people, three bedrooms, repeated street after street, in twenty-one suburbs.

The spare bedrooms are not the residue of families that shrank. They are the result of a choice about what kind of house to retire into.

What this means if you are buying

Three things follow, and none of them are advice about whether to buy.

Check the over-65 share before you commit, in either direction. In the 21 suburbs above, two in five residents are past retirement age against 17% for the country. If you have primary-age children that is worth knowing in advance — it bears on how many other families are around, and on the size of the local school cohort. If you are downsizing it is the same number read the other way, and it is the reason to look there rather than a reason not to.

Bedrooms per resident is not the same as houses being big. If you are comparing two suburbs on space, persons per bedroom will tell you where households are small relative to their homes. It will not tell you the houses are larger, because across the country they are not. Compare dwelling size separately or you will read one as the other — we nearly did.

You are bidding against retirees for three-bedroom houses. This is the practical consequence of everything above. In these suburbs a family-sized house is not priced by families; it is priced by couples who have sold somewhere else and do not need a fourth bedroom or a school catchment. If you are a young family looking at the same stock, that is who is on the other side of the auction, and they are frequently not borrowing.

What we measured, and what we left out

We could not check the report's own housing claim. The Intergenerational Report's most quoted number is home ownership among 25 to 34 year olds. The Census does not publish tenure crossed with age at suburb level — tenure comes broken down by dwelling structure only. We can tell you what share of a suburb owns its homes. We cannot tell you what share of its under-35s do, and we are not going to estimate it and present the estimate as a measurement.

We used the ABS figure, not our own arithmetic, for the suburb list. Average persons per bedroom is published by the ABS and rounded to one decimal place. That rounding is why the list above is a band — 21 suburbs that all report 0.6 — rather than a ranking from first to twenty-first. We can derive bedrooms per home by dividing household size by persons per bedroom, and we did so for the quintile table, where averaging hundreds of suburbs cancels the rounding out. We did not use that derivation to rank individual suburbs, because at one decimal place a single suburb's derived bedroom count carries a band of roughly ±8% and any ranking built on it would be mostly noise. As a check, the same derivation reproduces 3.18 bedrooms per dwelling nationally against the ABS published figure of about 3.1.

Almost everything here is measured within the 2021 Census alone. The suburb list, the bedroom gradient and the migration figures all come from a single Census, so suburb boundary changes between editions cannot distort them. Where we do compare 2016 with 2021 — the national over-65 shift, the ownership totals — the figures are summed across all 14,392 suburbs we could match, because a boundary moved between two suburbs cancels out in a national total. We do not publish suburb-level 2016-to-2021 changes here. Without a population guard the fastest-changing suburb in Australia looks like Park Ridge in Queensland, which appears to have got 25 percentage points younger — in reality its population grew 238% because an estate was built on it.

We are not publishing a list of suburbs that lost young people. The aggregate is real: among suburbs whose population barely moved between the two Censuses, 64.2% lost share among 25 to 34 year olds, a median of 0.44 points, and we checked it is not simply a high-density effect. But the individual biggest falls are inner-city apartment suburbs, and the 2021 Census was taken in August 2021, during lockdowns and with the international border closed. Naming those suburbs would be reporting the pandemic as though it were a trend.

The under-occupancy figure counts people, not need. "Bedrooms minus people" treats a couple sharing one room as having a spare. The occupancy standard the ABS uses for overcrowding does not. Our figure is a comparison between suburbs, not a count of empty rooms in any particular house.

What the first draft got wrong. It argued these suburbs aged in place, on two pieces of evidence that do not carry the claim. A stable population rules out growth and decline but not one household replacing another. A flat national ownership rate says nothing at all about turnover, because the split between owning and renting can sit still while every house on a street changes hands. Measuring migration directly answered the question, and answered it the other way.

Check we are measuring the right thing. Across the 14,392 suburbs we could match to both Censuses, our figures give a population of 25.34 million against the ABS count of 25.42 million — the gap being suburbs we could not pair — and a 75-and-over share of 7.55% against the ABS published 7.5%.

The data behind this

Every figure here comes from the 2021 and 2016 Censuses at Suburb and Locality level. The 2021 Census is the most recent count available; the 2026 Census will land directly into this debate, and we intend to run these same measurements against it when it does.