The public sector started 1.76% of the dwellings begun in Australia in the March quarter of 2026, and finished 2.58% of those completed. Across the second half of the 1980s the figure averaged 8.97%: about one dwelling in eleven. Across the 2020s so far it has averaged 1.93%, about one in fifty. In the last four quarters, 4,356 dwellings were begun by the public sector, out of 197,456 begun in total.
The node reading is measured against a 2011–2025 baseline, a period in which public delivery was already near 2 per cent. Measured against the full series back to 1984, when it ran between 5 and 15 per cent, the same reading sits in the lower third.
What counts as a public dwelling, and what does not
The ABS splits dwelling building by sector, and the split follows the intended owner of the completed dwelling, not who does the work or who pays for it. A dwelling built by a private builder under contract to a state housing authority counts as public, because the authority will own it. A dwelling built by a community housing provider may count as private, because the provider is not a government body.
This is therefore not a measure of social housing. It does not track social, affordable or community housing as those terms are used in policy, and a change in how governments deliver housing — through providers rather than directly, or the reverse — moves this series without any change in the number of subsidised homes. No series on social or affordable housing delivery is loaded in the platform: on that question the return here is NIL RETURN, and this piece does not stand in for it.
One further mechanic. The ABS publishes a total and a private-sector figure; the public figure used here is the difference between them. It is a residual, so it carries the rounding and revision of both.
Forty years of decline
In the September quarter of 1986 the public sector started 14.8% of all dwellings begun in Australia, the highest in the series. In the December quarter of 2015 it started 0.82%, the lowest. The path between the two is a long decline with one interruption.
By decade averages, the public share of commencements fell from 8.97% in the second half of the 1980s to 5.22% in the 1990s, 2.62% in the 2000s, 2.44% in the 2010s and 1.93% so far in the 2020s. In dwellings, the public sector began 69,808 homes in those five and a half years of the 1980s and 22,610 in the six and a quarter years of the 2020s, while total dwelling starts rose from 797,571 to 1,172,808.
Public-sector share of dwelling commencements and completions, per cent, September quarter 1984 to March quarter 2026
Source: ABS Building Activity, Australia, tables 33 and 37, number of dwelling unit commencements and completions by sector, original. The public-sector figure is total sectors less the private sector.
| Period | Average public share of commencements | Public dwellings started | All dwellings started | Public share of the period |
|---|---|---|---|---|
| 1984–89 | 8.97% | 69,808 | 797,571 | 8.75% |
| 1990s | 5.22% | 75,938 | 1,478,594 | 5.14% |
| 2000s | 2.62% | 40,149 | 1,553,662 | 2.58% |
| 2010s | 2.44% | 44,967 | 1,936,343 | 2.32% |
| 2020s | 1.93% | 22,610 | 1,172,808 | 1.93% |
Source: as above, with dwelling counts from the same tables. The 1980s row covers the September quarter 1984 to the December quarter 1989; the 2020s row runs to the March quarter 2026.
The one interruption
Public commencements lifted sharply in 2010. The share reached 12.4% in the March quarter and 11.7% in the June quarter, and 14,940 public dwellings were begun across the four quarters from September 2009 — 3.4 times the number begun in the four quarters to March 2026, and two-thirds of everything the public sector has begun in the 2020s. The June quarter of 2010, at 5,487 dwellings, is the largest quarter in the series.
This co-timed with the Commonwealth’s Nation Building Economic Stimulus Plan, whose social housing component was budgeted at $6.4 billion for about 20,000 dwellings. The series cannot attribute the lift to that programme, and it does not need to: what it shows is the shape, a two-year pulse that ended as quickly as it began. By the March quarter of 2012 the share was back to 1.53%.
The pulse also shows the lag between starting and finishing. The commencement share reached its stimulus-era high in the March quarter of 2010; the completion share followed three quarters later, at 9.8% in the December quarter.
Public-sector dwelling commencements, four-quarter total, dwellings, 1985 to March 2026
Source: ABS Building Activity, Australia, table 33. Each point is the sum of the four quarters ending in that quarter, computed as the public-sector share applied to total commencements.
Average public-sector share of dwelling commencements by period, per cent
Source: as above. The note against each bar is the number of public-sector dwellings started in that period.
Where the reading sits
APN reads this as Government-Owned Dwelling Delivery (21381), the equally weighted mean of the commencement and completion shares. At the March quarter of 2026 it stands at +0.1246σ, a position of 29.5 ranked against all 167 quarters of the series.
Those two figures need reading together. The reading is measured against a 2011–2025 baseline, a window in which public delivery was already close to 2% of dwellings, so a figure barely above zero means only that this quarter looks like the recent past. The position, ranked against the whole series, is the more informative of the two: even a quarter that is unremarkable by recent standards sits in the lower third of the last four decades.
The recent direction is upward, modestly. The eight quarters to March 2026 carried 8,645 public dwelling starts, the most in any eight-quarter run since the one ending in June 2012, which still contained stimulus-era dwellings. That is a rise from a low base: every eight-quarter run of the 1980s and 1990s was larger.
What the measure does and does not show
Beyond the sector definition above, three limits matter. The figures are original, not seasonally adjusted, and public numbers are small enough that a single large project moves a quarter. The public dwelling counts here are derived by applying the published share to total commencements, so they inherit any revision to either. And commencements are not completions: dwellings begun in a quarter may finish two or more years later, which is why the two lines in the first chart move together only with a lag.
The series also says nothing about what the dwellings are for, who lives in them or what they cost. It counts dwellings by intended owner and nothing else.
Findings are presented on the basis of data and evidence alone.
