Approving more has stopped building more
Australian apartment completions have barely moved in four years, no matter what enters the pipeline. The constraint has shifted from approvals to the capacity to finish — and almost no policy is aimed at it.
Apartment completions are pinned. Since 2022 they have averaged 61,323 a year and varied by only 4.5%. The starts feeding them varied 11.6% — 2.6 times as much. Output is flat while input swings.
The surplus is piling up on site. Apartments are being started at 83,256 a year and finished at 63,696, so 19,560 dwellings a year join a queue rather than a street. 151,902 are under construction now.
The queue is lengthening, not clearing. It takes 2.4 years of work at the current finishing rate, against 1.6 five years ago. On current flows it grows another 0.31 years longer every year.
01 · The ceiling
Four years of flat output against a decade of policy aimed at approvals
Australia completed 105,565 apartments in the year to September 2017. It has not come within five per cent of that figure in any quarter since. Today's rate is 63,696 — 60% of the record.
What makes this a ceiling rather than a trough is the stability. Across the past four years apartment completions have a standard deviation of 2,777 on a mean of 61,323: a band of 4.5%. Over exactly the same window the commencements feeding them swung 11.6%. The industry's ability to start apartments varies. Its ability to finish them does not.
02 · Where the difference goes
Nineteen thousand dwellings a year join a queue instead of a street
If more goes in than comes out, the difference has to sit somewhere, and it sits on site. 151,902 apartments are under construction — within 4% of the all-time peak — while commencements run at 71% of theirs.
That combination is the whole argument. A near-record amount of work in progress produced by well-below-record starts can only mean the work is moving through more slowly. The share of the on-site stock completed each year has fallen from 93% in 2006 to 42% now.
Detached housing has no such problem. Houses are started at 113,246 a year and finished at 109,074 — near enough the same number — and their queue is 0.8 years against the apartment sector's 2.4. Whatever is binding, it binds the form of housing that cities need most and the form that takes longest to build.
03 · Why it matters now
The national target fails at construction, not at approval
The National Housing Accord needs 68,643 completions a quarter from here. The best quarter Australia has ever recorded is 58,787. The gap is usually discussed as a planning problem — too few approvals, too slow a system.
The apartment data says otherwise. Starts are rising. Approvals are rising. What is not rising is finished dwellings, and the thing standing between the two is a construction sector that has completed roughly 61,323 apartments a year through a pandemic, a cost shock, a rate cycle and a migration surge without meaningfully deviating.
More approvals do not fix that. They lengthen the queue.
04 · Where it bites hardest
The queue tracks the apartment share, and NSW carries both
NSW has 1.8 years of work in the ground and 52% of its pipeline in apartments. TAS has 1.0 years and 7%. The ordering is not about planning competence; it is about what each state is trying to build.
Method
- Source. ABS Building Activity, quarterly, retrieved through the ABS Data API at vintage 20261001T002609Z. Six published figures are frozen in the repository and re-checked on every build; if ABS revises them the build fails rather than publishing quietly.
- Definitions. Started, finished and on-site are the ABS commenced, completed and under-construction series for "other residential", which is everything that is not a detached house. Annual rates are rolling four-quarter sums; the queue is the under-construction stock divided by completions over the prior year.
- The ceiling. Measured as the standard deviation of four-quarter completions since January 2022 over their mean, against the same statistic for commencements on the identical window.
What this does not show
- It does not identify the constraint. Flat completions against variable starts is consistent with a labour shortage, a materials constraint, insolvencies among builders, financing conditions, or projects deliberately paused. This data cannot separate them, and anyone claiming to know which it is from these series alone is guessing.
- Four years is a short window. The stability since 2022 is striking but it is sixteen observations of a series that was far more variable before. A ceiling that has held for four years may be a ceiling or may be a plateau.
- Some accumulation is normal. A pipeline in which starts exceed completions is growing, which is what you want when demand is rising. The concern is the combination: a near-record on-site stock produced by well-below-record starts, which means slower throughput rather than more building.
General information only. Nothing here is financial, investment or property advice.