When the Commonwealth gives money to the states for housing, the question that decides whether the policy worked is additionality: did this produce dwellings that would not otherwise exist, or did it substitute for spending a state was going to make anyway?

That question has an answering mechanism and the answers are public. Under the Accelerator, states and territories provide six-monthly Statements of Assurance from 2024 to 2028, signed off at deputy secretary level, and Treasury publishes them. Each carries total dwellings committed, commenced and completed, people or households housed, funding committed but not spent, and funding spent, alongside counting rules preventing a dwelling being reported twice at different stages.

What stands between a reader and the answer is now only shape. The statements arrive as combined PDFs, one file per reporting round with every jurisdiction inside it, so building the series any assessment of additionality requires means transcribing tables out of documents released months apart.

Read the combined Statement of Assurance PDFs on the Treasury website, one reporting round at a time, and transcribe. Treasury also publishes per-state Implementation Plans and periodic national completion counts, which give a cross-check on the totals.

Marked verified on 2026-09-03: the August 2024 combined statements were downloaded and read. Separate signed statements for six jurisdictions, each with tables for dwellings committed, commenced and completed, households housed, and funding committed against funding spent.

Checked on 2026-09-01, and the check overturned this entry.

The earlier version said the statements were written, sent to the Commonwealth and not published, and recorded not_published as the blocker. That was wrong. Treasury publishes them. The August 2024 round was downloaded and read: a single combined PDF running to roughly 2,600 lines of extracted text, containing separate signed statements for New South Wales, Victoria, Queensland, Western Australia, South Australia and Tasmania, each covering 1 January to 30 June 2024 with the dwelling and funding tables described above.

The correction matters beyond this entry, because this was the case used to argue that not_published belonged with the publishing choices rather than the constraints on the world. The argument itself still stands on its own terms -- the test is whether answering requires anyone to collect something new, and withholding an existing document does not -- but it should be understood as resting on reasoning rather than on this example, which turned out not to be an example of withholding at all. The reasoning is set out on the methodology page.

There is a lesson in how this entry came to be wrong. It was written from a survey of secondary reporting in which nobody mentioned the statements being public, and absence of mention was read as absence of publication. A register that makes that inference routinely will fill up with false gaps, which are worse than no entry.

The residual gap is real but modest, and the entry has been rewritten to claim only that: the statements are published in a form that cannot be queried, and stale is recorded because a six-monthly PDF released some months after the period it covers is a slow way to follow a program running to 2028.

One clarification the register should keep. The Accelerator was $2 billion, announced June 2023. The $3 billion often attributed to it is the separate New Homes Bonus. A third instrument, the National Agreement on Social Housing and Homelessness, carries $9.3 billion over five years from July 2024.

Related: How many homes has the Housing Australia Future Fund actually finished, and where are they?, How many social housing dwellings are vacant, and for how long, by area?.