Collected 3–4 October 2026. The completed work is a national funding dossier with selected entity and asset evidence. It is not a complete register of every tax law, transaction or ultimate owner in Australia.
What is available and what this package contains
| Public source | What it can answer | Completed collection and its limit |
|---|---|---|
| Final Budget Outcome | What the Commonwealth actually received, spent, owned and owed | Preserved the 2025–26 report; extracted receipts, the annual cash/financing statement, GST settlement and payments to each state. Reconciled deficits, financing and through-state/local exclusions. Not an agency-by-agency spending ledger. |
| ABS Government Finance Statistics | Consolidated revenue and expenses across governments | Preserved five workbooks covering the Commonwealth, states in aggregate, councils in aggregate, universities and consolidated government. Extracted operating accounts and reconciled the cross-level revenue difference. Individual state/council non-tax accounts and matched internal transactions remain separate work. |
| ABS Taxation Revenue | Tax collections by statistical category and jurisdiction | Preserved all workbook history; extracted 327 observations for 2024–25 across ten tables. Rows include subtotals and cannot all be added. |
| ATO corporate transparency | Named large entities' income, taxable income, income tax payable and PRRT | Preserved the 2024–25 release and extracted every income-tax and PRRT row, including separately identified late prior-year returns. It is not every business or every corporate-group payment. |
| DFAT trade statistics | Leading customers, suppliers and export/import categories | Preserved the calendar-2025 summary PDF and reviewed its tables and notes. Full commodity-by-country histories and confidential country allocations are not collected. |
| ABS international investment and DFAT investment | Country positions, investment transactions and income | Preserved the release/methods and extracted 65 rows from five selected ABS country tables. Immediate counterparties do not establish ultimate owners. |
| ABS balance of payments | Recent trade, primary income and current-account flows | Preserved the June-quarter release; reviewed the current-account components. Quarterly seasonally adjusted data stays separate from annual trade tables. |
| AOFM investor analysis | Broad classes of Commonwealth bondholders | Preserved the May 2026 article about December 2025 holdings. No complete named beneficial-holder list. |
| APRA superannuation | Scale of domestic retirement savings | Preserved June 2026 release. Not a look-through portfolio of Australian assets. |
| Commonwealth Grants Commission | How GST distribution is recommended | Preserved 2026 Update page and selected shares. Recommendations are not final-year cash receipts. |
| Treasury tax expenditures and ATO tax gaps | Concessions, benchmark choices and estimated non-compliance | Preserved the TEIS report and tax-gap index; explained the concepts. Individual concession estimates and tax-gap series are not extracted or independently assessed. |
| WA taxes and royalties | Detailed tax families, rates, exemptions and interstate comparisons | Preserved the 2025–26 handbook and resource revenue page. It is not a verified current statute register for every jurisdiction. |
| AusTender and GrantConnect | Which suppliers or recipients have reported contracts and awards | Preserved explanatory pages. No complete company-level contracts/grants join is presented. Awards do not measure actual annual payments. |
| AEC transparency register | Disclosed political financial relationships | Preserved the entry page; no donor-level extraction or causal influence finding. Political funding is a separate account from government revenue. |
| Issuer annual reports and ownership pages | Taxes, royalties, projects, legal owners and asset interests | Reviewed BHP, Rio Tinto, Woodside, Ausgrid, Port of Melbourne, Transgrid and Temasek examples. Self-reported contribution figures require perimeter and accounting checks. |
The download manifest records 52 local snapshots, totalling 26,651,973 bytes (26.7 MB), plus six failed retrieval attempts. A downloaded Treasury consultation page is a JavaScript shell and supplies no usable substantive text; it is explicitly excluded as evidence. Download counts are not counts of fully reviewed publications. The source notes identify extraction sources, context snapshots, failures and supplementary web-only references. These records have a separate namespace from the project's earlier source-family inventory.
Where public visibility stops
The foreign-ownership register is not a public directory. Treasury's Guidance Note 15, version 7, page 1 and section I expressly says its information is not public and is governed by protected-information rules. This source was reviewed through web text; it is not included in the local download count. Official reporting obligations do not grant us access to confidential submissions.
Corporate tax disclosure supplies selected fields, not entire tax returns. Country investment data can suppress sensitive details and identify an intermediate custodian or fund domicile. Listed-company substantial-holder disclosures, annual reports and asset owner pages can reveal more, but cannot be assumed to expose every beneficial investor. Public asset and company identifiers also change over time.
Public country-by-country reporting guidance provides another route for eligible multinationals, for reporting periods beginning on or after 1 July 2024. The guidance says publication will be through data.gov.au. This package has preserved the guidance; it has not verified a complete available company filing population or extracted those filings. Do not turn the existence of a reporting regime into a claim that every multinational's current data is already available.
No paid company extracts, accounts, registrations, FOI applications or outreach were used. Some needed answers may be obtainable through those channels; others may remain protected. A missing record is neither a zero amount nor evidence of misconduct.
Investigations that would add the most explanatory value
These are specific extensions of the completed national research, ordered by their ability to connect money to public value.
- Follow a gas export project from sale to public return. Identify the project, operator and partners; match production, sales, ownership, royalties, PRRT, company tax, grants, infrastructure commitments and rehabilitation obligations for the same period. Begin with a disclosed project such as Australia Pacific LNG, then verify the legal entities rather than assuming the operator's accounts cover every partner. The deliverable should reconcile what is known and leave confidential tax inputs visibly unknown.
- Build the major-business account across time. Start with the 25 disclosed entities in this dossier, map their corporate groups and collect prior annual tax files. Separate Australian and global revenues; compare tax payable, reported cash tax and royalties; then connect public contracts and grants by legal entity and date. This allows a meaningful explanation of changes rather than a one-year league table.
- Map strategic asset ownership. Expand the three infrastructure examples to electricity networks, ports, airports, telecommunications, water and major resource projects. Record effective dates, legal interests, managers, ultimate owners where evidenced and government ownership separately. Show acquisition proceeds separately from subsequent investment and annual public receipts.
- Complete the jurisdiction tax register. For every state and territory, collect current official revenue-office guidance, budget revenue statements and enabling law. Add council rates and material local charges with coverage limits. Each entry needs tax base, liable party, collector, rate/threshold, commencement, exemptions, revenue period and source. The current 327 statistical observations are the revenue scaffold for this work, not the completed legal register.
- Follow public spending back to businesses. Join contracts, grants, loans, equity investments and guarantees to recipients. Preserve commitments, payments, balances and contingent liabilities as distinct fields. Use agency accounts and audit reports to reconcile actual cash and delivery before calculating any company's net public contribution.
- Test national exposure to concentrated customers and suppliers. Extend commodity-by-country histories to prices, volumes, domestic value added and critical imported inputs. Distinguish export market concentration from an assertion of political control. Any shock model needs substitutability, contracts, inventories and behavioural assumptions; the current tables establish exposure, not the cost of a disruption.
Political donations and lobbying can be connected to those records as a separate investigation. The required evidence is a dated decision trail and the relevant legal/institutional context; a financial relationship alone does not prove a decision was bought.
Reproduce and inspect the evidence
From the repository root, using Python 3.11+ and the packages in the existing pinned research requirements:
python3 research/briefs/australia-money-flows/extract.py
python3 research/briefs/australia-money-flows/verify_fiscal.py
Both commands default to read-only. The original extractor checks preserved snapshot SHA-256 hashes, source headings/periods, entity uniqueness and population, aggregate controls, selected arithmetic and exact reproduction of the saved derived JSON. --write regenerates the derived files and validation report; it does not change the original publications. The check details are in validation.json. The fiscal verifier separately reproduces the operating, cash, borrowing, GST and per-state transfer account, with its check details embedded in fiscal-reconciliation.json. Its --write flag regenerates that ledger. Matching a publisher's totals does not independently audit the underlying reporting.
| File | Contents |
|---|---|
| corporate-entities.json | 4,415 income-tax records with preserved blank fields, plus 21 PRRT records |
| corporate-summary.json | Current-year controls and the top 25 entities |
| tax-observations.json | 327 tax-category observations with period, basis and workbook locator |
| government-observations.json | Consolidated government revenue and expense purpose categories |
| fiscal-reconciliation.json | Five ABS operating accounts, Commonwealth annual cash/financing, GST settlement and per-state transfers; includes overlapping totals |
| commonwealth-receipts.json | FBO 2025–26 actual cash-receipt rows, including subtotals and memoranda |
| international-observations.json | Selected investment stock, transaction and income country tables |
DFAT's trade table, FBO receipts and Transgrid's ownership chart were also visually inspected. Other numerical claims identify their source in the prose; not every narrative statement is covered by an automated check.
Originals are preserved locally under raw/ and are git-ignored, following the project's existing research-library practice. A clean checkout needs the exact originals obtained from the URLs in the manifest; save each at its recorded path and verify its SHA-256. The earlier GFS workbook was copied byte-for-byte from the existing local evidence collection, with that provenance recorded. Publisher pages may change, so a fresh download may not reproduce the saved hash. Such a mismatch requires an explicit source-vintage update, not silently changing the hash to make the check pass. There is no automatic collector or scheduled refresh in this package.
This manifest records requested/resolved URLs, retrieval times for successful snapshots, lengths, content types and hashes. The four ABS workbooks added on 4 October also preserve full HTTP response headers and their hashes; earlier entries do not. Failed entries preserve the error but lack a per-attempt timestamp; they belong to this 3 October collection session. BHP and Temasek direct downloads were blocked by HTTP 403; selected cited content was available through web retrieval. An RBA direct request failed certificate validation; verification was not disabled. Two superseded ATO URLs returned 404 and were replaced with working sources. These limitations remain in the log.
Attribution and future updates
The research draws on Australian Bureau of Statistics, Commonwealth of Australia agencies, state government and named corporate publications. Original publications retain their own licences, third-party exceptions and attribution requirements. They are local study copies, not MIT-licensed project code or a cleared public mirror. Link readers to publisher originals. Company disclosures remain issuer claims unless separately corroborated.
For each update, retain the publisher, requested and resolved URL, publication/revision date where known, retrieval date, observation period, currency/unit, entity scope, accounting basis, original hash and precise table/page/cell. Label actuals, estimates, commitments and calculated values separately. Keep revised vintages alongside the old evidence when they change a conclusion. Verify the source and calculation before moving a figure into app data or a policy model.