The Downsizing Dead End That Isn't

This piece exists to report a hypothesis that failed. The premise was tidy: Australia’s outright-owning older households — the cohort It’s Not Wealth. It’s Time. showed dominates the nation’s oldest local government areas — are locked in place partly because new supply is not built where they live, mechanically trapping the largest dwellings under the smallest households. A “downsizing dead end,” measurable as a strong negative relationship between an LGA’s over-65 share and its building intensity. Pre-registered, then tested against the certified tables. The data declined to provide it, twice over — and both refusals teach something the premise obscured, including the identity of an unexpected construction engine: the retiree coast.

Refusal one: the relationship is weak, and it bends

Across 364 LGAs with adequate sample size, the correlation between over-65 population share and two-year approval intensity is −0.166 — the right sign for the hypothesis, but a fraction of the strength that would support a “systematic starvation” claim (compare the +0.884 age–ownership relationship, or the +0.42 mortgage–intensity one). And the quintile pattern is not even monotonic. Building intensity falls steadily across the first four age quintiles — 4.69 per cent of the dwelling base in the youngest fifth of LGAs, down to 2.72 per cent in the fourth — and then rises to 3.14 per cent in the oldest fifth. The oldest council areas in Australia are out-building the merely old ones.

The full picture: a cloud, not a line

Figure 2 shows every one of the 364 councils, per the house standard of publishing the full sample rather than a curated exhibit. The honest description of this chart is that it is a cloud. High-building councils exist at every age share from 3 per cent over-65 to 42; near-zero builders exist across the same span; the faint downward drift that produces r = −0.166 is visible only if you look for it. If the premise were true — if ageing areas were systematically starved of construction — this chart would show a wall falling left to right. It shows weather.

The sea-change engine

The uptick in the oldest quintile has names, and they map a recognisable coastline and river system. Victor Harbor — at 42.2 per cent over-65 the oldest council in the sample — carries a two-year approval intensity of 9.88 per cent, placing it among the most intensive builders in the country relative to its size. Alexandrina (32.7 per cent over-65) builds at 7.46; Copper Coast at 5.74; Fraser Coast in Queensland at 7.18; Mid Murray at 8.79; Murray River at 6.05 (Figure 3). These are Australia’s retirement destinations, and they are active construction markets because they are old: supply follows the retiree, arriving in volume where downsizers move to. What the premise imagined as a cohort abandoned by the construction industry turns out, at the oldest end, to be the construction industry’s customer.

Refusal two: the elderly majority does not live where the premise assumed

The sharper correction comes from weighting by people instead of places (Figure 4). Of the roughly 4.3 million over-65s living in the sampled LGAs, some 55 per cent live in the two youngest, highest-building quintiles — the metropolitan council areas — and only about 13 per cent live in the oldest fifth. The mental image behind the hypothesis (the typical older Australian marooned in a low-supply, demographically aged area) inverts the actual distribution: the typical older Australian lives in a big-city LGA where building intensity is the highest in the country. Whatever obstructs downsizing at scale, it is not, on this evidence, the wholesale absence of local construction activity.

What the certified data cannot yet answer

One honest limitation prevents this null from being the final word. The certified approvals table counts dwellings; it does not classify them. Intensity here includes detached houses on greenfield lots — supply, but not downsizing supply. The precise question the premise wanted asked is whether suitable stock (smaller dwellings, single-level, low-maintenance) is being added where downsizing demand originates, and that requires a dwelling-type split the certified tables do not currently carry. That series is registered as future acquisition work; until it exists, the type-level claim is a NIL RETURN, and this analysis says so rather than proxying around it. What can be said on the certified data is narrower and still useful: the geography of ageing and the geography of building are only weakly opposed; the oldest markets actively build, led by the retirement coast itself; and most older Australians live amid the country’s heaviest construction. If a downsizing dead end exists — and the national picture of large dwellings under small households says something does — it is a product-mix problem, not a location problem, and the policy conversation should move accordingly.