INSIGHT · RISK & FINANCIAL VISIBILITY

One source, two numbers.
A construction insolvency count
needs its cutoff.

26 August 2026 · Updated 25 September 2026

How many Australian construction companies entered insolvency for the first time in FY2025-26? On ASIC's Series 1 workbook published 24 August 2026, with data to 9 August 2026, the answer is 3,472 - 3.4% below FY2024-25. Industry coverage in July 2026 reported a lower figure from an earlier weekly release. Neither reading is an error. An insolvency count is a number plus a series and a cutoff date, and quoting one without the other two chooses a date on a reader's behalf.

Two construction professionals in hard hats comparing figures on a tablet at a building site

Analysed 26 August 2026 · ASIC weekly releases 13 July 2026 → 24 August 2026 (data to 28 June and 9 August 2026) · re-read 25 September 2026 against the workbook published 21 September 2026 (data to 6 September 2026) and the Series 3 set released December 2025.

Key takeaways

The current reading is 3,472: Construction division, ASIC Series 1, FY2025-26, data to 9 August 2026, published 24 August 2026 - recomputed from the workbook's 72,453 record-level entries, matching ASIC's summary tables year for year.
The July reading was 3,435, and it carries lower confidence: Coverage published 20 July 2026 quoted it for the same series and year, citing ASIC's 13 July release with data to 28 June - two days before the financial year closed. We could not obtain that weekly file, so the value is as transcribed; the roughly 37-company difference is a transcribed difference, not a definitional one.
The gap is timing, not definition: Both readings use the same Series 1 basis. If the difference were about excluding controller appointments, it would be 299 companies, not 37.
The larger definitional choice sits in the same file: For the same industry, year and cutoff, Series 1 records 3,472 companies and Series 2 records 4,439 appointments - a gap of 967, or 27.8%.
The decline is not a relief signal: Entries via the small business restructuring pathway fell 335, while court-ordered liquidations rose 157 (+23%) and receiver appointments rose 53 (+73%).
Construction's position is unchanged: 3,472 first-time events against 2,078 for the next-ranked division, and 24.5% of the whole-economy total (24.4% a year earlier).

What changed

Between 13 July and 24 August 2026, ASIC published two weekly versions of the same statistic, and the FY2025-26 construction figure was not the same in both. The July version was picked up by coverage that read the year as a fall; separate coverage two days later described whole-economy insolvencies as climbing. Two narratives sat side by side in the same fortnight, drawn from the same register.

We downloaded and archived the Series 1 and Series 2 workbooks published 24 August 2026 and 21 September 2026, and recomputed the figures on this screen from their record-level entries - appointment type, effective date, industry division, state and series flags - cross-checking the aggregation against ASIC's summary tables. Where a media figure and the workbook disagree, we say so rather than pick one.

Data & Methodology

Why the two readings differ -
and why the obvious explanation is wrong

A natural first guess is that the figures use different definitions - that one counts controller appointments and the other does not. The record-level data rules that out: both readings are Series 1.

ASIC says as much in its interpretation guidance: appointments less than a month old are "considered provisional", and "current statistics will not always reconcile with prior overlapping periods."

The arithmetic settles it: In FY2025-26, 299 construction companies had a controller-class appointment - receiver, receiver and manager, controller, or managing controller - as their first recorded event. Removing that class would move the count by 299, not by roughly 37.
The mechanism is more mundane: Administrators and controllers must lodge notice of appointment within a statutory period, and ASIC republishes weekly, rolling in lodgements as they arrive. The 13 July release closed its data at 28 June - two days before the financial year ended - so it did not yet cover the full year it was being read as. The 24 August release closed at 9 August, forty days after year end.
The real definitional choice is larger: For the same industry, year and cutoff, Series 1 records 3,472 companies while Series 2 - all appointments, including subsequent and transitional ones - records 4,439. A gap of 967, or 27.8%, between two figures a reader could reasonably call "ASIC's construction insolvency number". Which series you cite is a genuine methodological decision; it simply is not what separated the two headlines.

What the data shows

On a consistent basis - Series 1, construction division, every year read at the 9 August 2026 cutoff:

Source: ASIC Insolvency statistics, Series 1 and Series 2 workbook, published 24 August 2026, data to 9 August 2026. Construction division, Series 1, full statutory register (not a sample). Cyberate recomputation from 72,453 record-level entries; aggregation matches ASIC's published summary tables for each year. FY2025-26 remains provisional.

Financial year	First-time insolvency events (companies)	Year-on-year change
FY2021-22	1,284	-
FY2022-23	2,213	+72.4%
FY2023-24	2,977	+34.5%
FY2024-25	3,596	+20.8%
FY2025-26	3,472	−3.4%

Two findings, computed at the 9 August cutoff: Construction remains the insolvency-heaviest division - 3,472 first-time events against 2,078 for the next-ranked division, accommodation and food services - a 67% lead. And construction's share of the whole-economy total is essentially unchanged at 24.5% (3,472 of 14,153), against 24.4% (3,596 of 14,722) a year earlier. The level moved; the position did not.
One headline we cannot reconcile, and will not pick a side on: July coverage described whole-economy insolvencies as climbing for FY2025-26. On the archived workbook, whole-economy Series 1 is 14,153 against 14,722 the year before - a fall of 3.9%. The comparison basis behind "climbing" is not stated in that coverage, so it cannot be checked, and we do not assume it is wrong. A direction claim without its series, period and cutoff is not verifiable by the reader, only repeatable.

What's driving the decline -
and why it isn't a relief signal

A net figure of −124 companies invites a single story: pressure easing. The composition tells a different one. Broken down by the type of each company's first recorded event, the year-on-year change accounts fully for the −124:

Source: Cyberate recomputation, ASIC Series 1, construction division, by first recorded appointment type. All categories shown; columns sum to the totals given.

First event type	FY2024-25	FY2025-26	Change
Restructuring (small business restructuring)	785	450	−335
Court liquidation	683	840	+157
Creditors' voluntary liquidation	1,579	1,631	+52
Receiver appointed	73	126	+53
Provisional liquidation	5	23	+18
Controller appointed	60	63	+3
Managing controller	0	1	+1
Receiver and manager	119	109	−10
Voluntary administration	292	229	−63
Total	3,596	3,472	−124

One line more than accounts for the net decline: Entries via the small business restructuring pathway fell 335. That pathway is sensitive to policy settings and adviser practice, so a fall in its use is not a clean signal that distress has receded - the same contraction appears economy-wide, where restructuring appointments dropped from 2,918 to 1,714 (−41.3%). Meanwhile the procedures no company chooses for itself moved the other way: court-ordered liquidations rose 23% and receiver appointments rose 73% from a small base.
None of this indicates where the register goes next: Insolvency statistics are a lagging record of distress that has already occurred, and this article makes no forecast. It does mean the shape of the decline and the direction of the headline point different ways, and a decision leaning on the headline alone has not read the data.

The cutoff argument is about weeks.
The lag underneath it is years.

ASIC answers a different question about the same register in a third series, and that answer is measured in years. Series 3 is built from the initial reports external administrators and receivers lodge after an appointment, and each report carries the administrator's own estimate of when the company actually became insolvent. For construction in FY2024-25, that estimate sits a long way behind the appointment date the register records.

Source: ASIC Insolvency statistics Series 3.2, external administrators' and receivers' reports for selected industries, 1 July 2024 to 30 June 2025, table 3.2.2.13, construction industry. Workbook stamped Released: December 2025; downloaded and read 25 September 2026. The population is the 1,944 construction reports that alleged insolvent trading and gave detail of when insolvency began, not all 2,361 construction reports lodged.

A companion table in the same workbook, table 3.2.2.3, flags 1,955 construction reports under section 588G insolvent trading; we quote 1,944 because that is the total the timing table itself carries. The all-industries comparison below is from a different table, Series 3.1 table 3.1.11.4, and its 86.3% is our own sum of the same three rows we sum for construction. Shares are ASIC's own column, rounded to one decimal.

When the company became insolvent	Construction reports	Share of reports
At appointment	15	0.8%
1 to 3 months before appointment	48	2.5%
4 to 9 months before appointment	189	9.7%
10 to 15 months before appointment	280	14.4%
16 to 24 months before appointment	341	17.5%
Over 2 years before appointment	1,071	55.1%
Total	1,944	100%

87.0% were already insolvent ten months or more before the appointment: 1,692 of the 1,944 construction reports put the onset of insolvency at ten months or more before the appointment the report was lodged against, and 1,071 of them put it more than two years before. The 1,692 is our own sum of three rows ASIC publishes separately, over ASIC's published total; all four numbers are in the table above. A dispute about a six-week cutoff is a rounding error against a lag of that size.
This is an estimate on a form, not a measurement: ASIC states that Series 3 financial information "reflects the estimates and opinions of the external administrator at a point in time", that revised information from later reports "is not reflected in these statistics", and that administrators "predominantly answer questions by selecting from a predetermined set of options and numerical ranges". The onset date is one administrator's judgement, recorded in bands, and a later revision to it never reaches this table.
Construction is not distinctive here: The equivalent table for all industries, Series 3.1 table 3.1.11.4, reads 55.7% over two years and 86.3% at ten months or more, on 7,749 reports. Construction sits just below the whole economy on the over-two-years band and just above it at ten months or more, which is a split rather than a gap in either direction. The lag is a property of what the register records, which is the appointment and not the insolvency, rather than a finding about this industry, and we are not claiming otherwise.

What Series 3 counts,
and what it leaves out.

Series 3 is a different population from Series 1, not a later view of the same one. Series 1 counts companies entering external administration. Series 3 counts reports, mostly the ones a lodgement obligation produced and some lodged voluntarily for statistical purposes.

ASIC says the two do not line up: the statistics "will not directly correlate with other insolvency statistics (Series 1 and Series 2) due to the lag in a report being lodged after the date of appointment and not all appointments requiring the lodgement of a report". For trends ASIC names its own preference, and it is the basis this article already used: "We prefer Series 1 statistics to be used when explaining trends in corporate business failures".

Sources: Series 3 report counts from ASIC Series 3.3, external administrators' and receivers' reports time series, 1 July 2004 to 30 June 2025, table 3.3.5, construction column; workbook stamped Released: December 2025, read 25 September 2026. The FY2024-25 count of 2,361 is the same figure in Series 3.1 table 3.1.1 and in the Series 3.2 construction sheet. The 238 and 247 report counts below, and the whole-economy total of 9,585, are from ASIC Series 3.1 table 3.1.1, same period and same December 2025 release.

Series 1 counts recomputed by Cyberate from the 74,627 data rows on the Data set sheet of the ASIC Series 1 and Series 2 workbook published 21 September 2026, data to 6 September 2026, each row carrying a Series 1 Yes or No flag. All five Series 1 values in the column above are identical to the ones this article published from the 24 August 2026 workbook, at a cutoff four weeks later, and unchanged. The share column is Cyberate dividing two counts ASIC publishes in separate workbooks; ASIC publishes no such ratio and warns the two series will not directly correlate.

Financial year	Series 3 construction reports	Series 1 construction companies	Series 3 as a share
FY2021-22	919	1,284	71.6%
FY2022-23	1,541	2,213	69.6%
FY2023-24	1,952	2,977	65.6%
FY2024-25	2,361	3,596	65.7%
FY2025-26	NOT PUBLISHED	3,472	-

There is no FY2025-26 Series 3, and 30 September is the date that would change that: As at 25 September 2026 the newest Series 3.1, 3.2 and 3.3 workbooks on ASIC's insolvency statistics page all stop at 30 June 2025. INFO 80 states that Series 3 "Excel workbooks are available up to 30 June of the most recent financial year (updated by 30 September each year)", so that is the release which would fill the blank row above. Set against it: the FY2024-25 set is stamped Released: December 2025. We are naming ASIC's stated cadence and ASIC's own date stamp, not predicting a release date, and the current INFO 80 text was itself updated in February 2026, so we cannot say what cadence applied to the last cycle.
Series 3 carries about two reports for every three Series 1 companies: FY2024-25 produced 2,361 construction reports against 3,596 construction companies in Series 1, a difference of 1,235. These are two counts of two populations over the same year, not one cohort followed through, because ASIC warns the lodgement lag means a report in one year can belong to an appointment in another. ASIC's lodgement trigger explains the shape of the gap: a report is required where it appears a relevant person "may have committed an offence in relation to the company, been negligent, or otherwise engaged in misconduct", or, in a liquidation, where the company "may be unable to pay its unsecured creditors more than 50 cents in the dollar". Some are lodged "for statistical purposes only" where neither applies. Economy-wide the same division gives 9,585 reports against 14,722 companies, 65.1%, beside construction's 65.7%.
The coverage share is drifting, and nothing published says why: Series 3 covered 71.6% of Series 1 construction companies in FY2021-22, 69.6% in FY2022-23, 65.6% in FY2023-24 and 65.7% in FY2024-25. That could be lodgement behaviour, the changing mix of appointment types, or something else entirely. Nothing in the three Series 3 workbooks or in INFO 80 establishes a cause, and we are not supplying one.
The regional cut exists, annually, and it is uneven: Series 3.1 and Series 3.2 break construction down by principal place of business, at state and territory level and no finer. In FY2024-25 the Series 3 share of Series 1 construction companies was 72.0% in New South Wales, 1,128 reports against 1,567 companies, and 53.7% in South Australia, 51 reports against 95 companies. South Australia's base is small enough that one year proves little, which is the same caution this article already attaches to its state figures. ASIC also states that Series 3 region comes from the "'principal place of business' recorded on our corporate register at the time the company entered into external administration", which is a different snapshot from the one behind the Series 1 state split above.
What this addition does not establish: It does not say construction insolvency is rising or falling: Series 3 is an annual count of reports lodged, and the newest year published is FY2024-25. It does not explain the coverage drift. It does not make the two series comparable, because every share in the table above is our division of two counts ASIC publishes separately and warns do not correlate, and the two industry shares that look alike, 24.6% of Series 3 reports against 24.4% of Series 1 companies, rest on that same uneven footing. It does not describe the trades as a reader might define them, since ASIC's own labels put "Construction (e.g. carpentry, plastering)" in one row and "Electricity, gas, water and waste services (e.g. electrical, plumbing)", 238 reports, in another, beside a "Labour hire" row of 247 that Series 1 and Series 2 do not carry at all. And it is not a forecast, not an assessment of any company, and not financial, legal, credit or investment advice.

What this means for customers

Multi-entity construction groups: If group consolidation or board reporting cites an industry insolvency benchmark, the series and the cutoff have to be agreed before the figure is used, not discovered afterwards. Series 1 against Series 2 is a 27.8% difference on the same cell; the same cell moved between two releases six weeks apart. A month-end number without a series label and a cutoff date is not a benchmark.
Suppliers and distributors extending trade credit: A monitoring basis that watches only "external administration" does not cover the 299 construction companies whose first recorded event in FY2025-26 was a controller-class appointment. The useful question is narrower - whether your data feed includes that class at all, or silently excludes it. This is a coverage question, not credit advice.
South Australian builders and developers: South Australia recorded 93 construction Series 1 events in FY2025-26 against 95 the year before - 2.7% of the national construction total, essentially flat. New South Wales alone accounts for 44.4% (1,540 events), so the national figure largely describes conditions in the largest states. A national narrative should not be assumed to describe South Australian exposure in either direction.

Where this sits in a system

This is the working logic we apply in the Group Finance & PM Platform: an external reference series is worth storing as an object carrying its four elements - period, basis, sample, cutoff - rather than as a bare number, so a consolidated report can state which ASIC release each figure came from.

What to watch next

The next ASIC weekly release: FY2025-26 is still maturing; anyone citing the figure should re-pull the latest workbook first. That the number hangs off a release date is the argument of this article, not a footnote to it.
Receiver and controller appointments: Construction: 252 → 299 year-on-year. Secured-creditor enforcement is a pressure signal worth tracking, but the sample is small and it supports directional observation only - never a conclusion about an individual company.
Restructuring pathway volumes: Construction −335; economy-wide −41.3%. Because this line is policy- and practice-sensitive, it is the most fragile leg of the "decline" narrative - watch whether it reverses as pathway usage shifts rather than as distress changes.
The mature FY2025-26 figure: Once late lodgements stabilise, the final reading against −3.4% is a natural audit of the cutoff argument itself.

Customer questions

Which figure should a group report cite - 3,435 or 3,472?

Whichever it cites, with its series and its cutoff attached. The defensible current citation is 3,472, ASIC Series 1, construction division, FY2025-26, data to 9 August 2026, published 24 August 2026, still provisional. The point is not that one figure won; it is that a figure without those labels cannot be checked by the person reading the report.

Does the fall mean construction distress is easing?

The file does not support that reading. The net fall of 124 is more than accounted for by a 335 drop in small business restructuring entries, while court liquidations rose 157 and receiver appointments rose 53. The direction of the headline and the composition beneath it point different ways.

Is this a forecast of where insolvencies go next?

No. Insolvency registration confirms distress that has already occurred. ASIC says the FY2025-26 figure may rise as late lodgements arrive, and that Series 3 workbooks are updated by 30 September each year. Whether the figure settles is something a later release answers, not this one.

Can an agent maintain this?

It can maintain the mechanical part - watching for the weekly release, re-pulling the workbook, attaching series, period, sample and cutoff to the figure, and drafting the note that says what changed. A person confirms and signs off before anything is used or sent.

Bring us the report
where two numbers disagreed.

The point of the 37-company gap is not which figure won. It is that a benchmark without its series and its cutoff cannot be checked by the person reading the report - and a month-end number without those labels is not a benchmark.

This article is general information about published regulatory statistics. It is not financial, legal, credit or investment advice, not a valuation, and not a forecast. It makes no assessment of any individual company or counterparty. All ASIC figures are as published in the release of 24 August 2026 with data to 9 August 2026 unless stated otherwise, and are provisional for recent periods.

Send us the report · Group Finance & PM Platform

Sources

Suggested citation: ASIC Insolvency statistics, Series 1 and Series 2 workbook, published 24 August 2026, data to 9 August 2026, archived and recomputed at record level by Cyberate. Current as at 25 September 2026: re-read against the workbook published 21 September 2026, data to 6 September 2026, and unchanged.

One source, two numbers. A construction insolvency count needs its cutoff.