In the spring of 2021, a small residential builder in one of the growth corridors on Melbourne's northern fringe signed nine fixed-price contracts. Each promised a four-bedroom house on a slab, for a set sum, by a set date, to a young family who had queued at a land release and stretched to the top of what a bank would lend them. The margin built into each job was thin — single figures, which is ordinary for volume residential work. By the time the frames went up in late 2022, the timber, the steel, the concrete, the trusses and every trade on site had repriced. His cost to finish those nine houses had risen by roughly a quarter. There was no clause in the contracts that let him pass any of it on. He built them anyway, because the alternative was to be sued for breach and lose his registration, and each completed house pushed him further into the red. The company that entered administration in 2024 had not lost a client, missed a milestone, or built a bad house. It had made one mistake: it signed a promise about 2023 prices using 2021 information.
Behind him stands a second figure who almost never appears in the coverage. The concreter who poured those slabs, the plumber who roughed them in, the truss supplier who delivered on thirty-day terms — when the head contractor collapses, those businesses are unsecured creditors. They rank behind the employees and behind the secured lender, and ASIC's own reporting on completed external administrations shows that unsecured creditors in Australian corporate insolvencies typically recover very little, with a large share of administrations returning nothing at all (approx., ASIC insolvency statistics, Series 3). The subcontractor does not merely lose a job. He loses work already done, materials already bought, and wages already paid to his own people. That is how one failure in a growth corridor becomes four or five, and why the number of company insolvencies is a poor measure of how much damage an insolvency wave actually does.
This topic is about the business side of the squeeze. A separate topic on this site — Anatomy of the cost-of-living crisis — traces what the inflation shock did to household budgets. What follows traces what the same shock, and the policy response to it, did to the balance sheets of the companies those households work for.
What happened: suppression, then resumption
The insolvency wave of 2023 to 2026 cannot be read without the two years that preceded it. Between March 2020 and the end of 2021, Australian governments did something unprecedented: they deliberately and temporarily switched off most of the machinery by which failing companies fail.
The suppression phase, 2020–2022
Three things happened at once. First, the Coronavirus Economic Response Package Omnibus Act 2020, passed in March 2020 under the then Coalition government, gave directors temporary relief from the insolvent-trading duty — the provision in the Corporations Act that makes a director personally liable for debts incurred while the company cannot pay its bills. It also lifted the minimum threshold for a statutory demand from $2,000 to $20,000 and stretched the time to respond from 21 days to six months. A statutory demand is the standard first step a creditor takes to force a company into liquidation; for most of 2020, that step was effectively unavailable for ordinary trade debts.
Second, JobKeeper — a wage subsidy of roughly $90 billion reaching around 3.8 million workers at its peak (approx., Treasury and ANAO reporting, 2020–21) — paid a large share of the wage bill for businesses whose turnover had collapsed. Wages are the cost that most often triggers a small business failure, and for a year that cost was substantially socialised.
Third, and least visible at the time, the Australian Taxation Office paused most of its firmer debt-recovery action. Director penalty notices, garnishee notices and wind-up applications largely stopped. Businesses were encouraged to defer PAYG withholding, GST and superannuation guarantee charge liabilities and to deal with them later. That pause was a deliberate policy choice, made for good reasons, and it worked in the narrow sense: company insolvencies fell to roughly 4,000 to 5,000 a year across 2020–21 and 2021–22, against a pre-pandemic run-rate of roughly 8,000 a year (approx., ASIC insolvency statistics, Series 1A). For two years, Australia recorded roughly half the business failures it would normally expect during a period of extraordinary economic disruption.
It matters for the fairness of this analysis that all three of those measures were designed and legislated under the previous Coalition government, and that JobKeeper had ended before the change of government in May 2022. The suppression phase was inherited, not caused, by the current federal government. What it inherited was not a clean slate. It was a stock of deferred failure and a very large stock of deferred tax debt.
The resumption phase, 2022–2026
The moratoria were always temporary and always intended to be. The insolvent-trading relief lapsed at the end of 2020; the statutory demand threshold was reset to a permanent $4,000 during 2021 (approx.); JobKeeper ended in March 2021. What followed under the current government was less a new policy than the switching back on of several old ones, close together in time.
The most consequential was the Tax Office's return to firmer debt collection. From 2022 the ATO signalled publicly, through commissioner-level speeches and its published debt-collection guidance, that the pandemic pause was over and that it would resume the enforcement tools it had shelved. In practice that meant three things: director penalty notices, which make a company's directors personally liable for unpaid PAYG withholding, GST and superannuation guarantee charge, issued again at scale — tens of thousands over the following years (approx., ATO annual reports and published debt-collection reporting); the disclosure of business tax debts above $100,000 and more than 90 days overdue to commercial credit reporting bureaux, a power legislated in 2019 but used sparingly until this period; and a return to court applications to wind up companies that would not engage, with the ATO re-emerging as one of the most active applicants for court-ordered winding up in the country (approx., ATO and ASIC published statistics).
Alongside that, the cost of carrying an ATO debt rose twice. The general interest charge, the penalty interest the ATO applies to overdue tax, sat above 11% a year and compounds daily (approx., ATO published GIC rates, 2025) — a rate set by formula rather than by ministerial decision, but one that tracks the cash rate and therefore rose with it. Then, from 1 July 2025, the general interest charge and the shortfall interest charge ceased to be tax-deductible, a measure announced in the December 2023 Mid-Year Economic and Fiscal Outlook and legislated in 2025. For a company on the standard 25% small-business rate, removing deductibility raises the effective after-tax cost of an ATO debt by roughly a third. The Tax Office had, for years, been the cheapest lender in the country to a business under strain. That was the point of the change, and it worked as designed.
Two other policy currents landed on the same balance sheets in the same window. The Reserve Bank of Australia, acting independently of government, lifted the cash rate from 0.10% to 4.35% across thirteen increases between May 2022 and November 2023 (RBA cash rate decisions). And the industrial relations framework was rewritten across three Acts — the Secure Jobs, Better Pay Act in 2022 and the two Closing Loopholes Acts in 2023 and 2024 — introducing changes to casual employment and conversion, labour-hire pay arrangements, multi-employer bargaining and, from August 2024 for larger employers and August 2025 for small business, the right to disconnect. Separately, the superannuation guarantee completed its long-legislated climb from 10% to 12% of ordinary earnings by 1 July 2025 — a schedule set in law well before this government took office, and therefore inherited rather than chosen.
Finally, one state-level rule deserves specific mention because it sits directly on the mechanism that failed. Victoria's Domestic Building Contracts Act 1995 restricts the use of cost-escalation, or "rise and fall", clauses in domestic building contracts below a prescribed value (approx. $500,000). The policy intent is sound and consumer-protective: a family building a home should not sign a contract whose price can be revised upward at the builder's discretion. The consequence, in an inflation shock, is that the entire risk of a materials and labour price spike sits on the builder's side of the contract, and the builder is almost always the smaller and thinner-capitalised party. That rule is a Victorian Labor inheritance rather than a Victorian Labor invention — it has been law since 1995 — but it was not revisited when the shock arrived.
The strongest case for the policy
Most of this wave is deferred failure, and business exit is not a policy failure
Insolvencies were artificially suppressed for two years. Insolvent-trading moratoria, a $20,000 statutory demand threshold, roughly $90 billion of JobKeeper and a near-total pause in ATO enforcement held company failures to roughly half their normal rate through 2020–21 and 2021–22 (approx., ASIC Series 1A). Businesses that would have failed across that period in any ordinary economy did not fail. That backlog did not vanish; it was pushed forward into 2023 to 2026. A large share of what is now called a "wave" is catch-up, and comparing today's total against the suppressed years overstates the deterioration substantially. Second, business exit is a normal and necessary feature of a functioning market economy. Capital and labour trapped in firms that cannot cover their costs are capital and labour unavailable to firms that can. A very low insolvency rate is not automatically a sign of health — sustained artificially, it produces zombie firms that survive on forbearance while absorbing skilled workers and credit. Third, the underlying construction problem originates offshore. Fixed-price contracts exposed to a global supply-chain and energy shock is a mechanism no Australian government designed, and the shock that detonated it — pandemic freight disruption, then a war-driven spike in energy and materials prices from February 2022 — was not caused here. And finally, the ATO cannot indefinitely decline to collect lawfully owed tax. Every dollar of PAYG withholding and GST sitting unremitted on a company's balance sheet is money already withheld from employees or already collected from customers on the Commonwealth's behalf. A tax office that permanently declines to collect it is not being kind; it is transferring a competitive advantage to the businesses that did not pay, at the direct expense of the ones that did. That is a serious equity problem, and resolving it was always going to be uncomfortable.
The evidence
The headline number is large and it is real. In FY2025–26, 14,152 companies entered external administration or had a controller appointed (ASIC insolvency statistics, FY2025–26). That is roughly three-quarters again the pre-pandemic run-rate of around 8,000 a year, and roughly three times the suppressed years of 2020–21 and 2021–22 (approx., ASIC Series 1A). It is among the highest annual totals ASIC has recorded in its published series.
The distribution is where the story lives. Construction accounted for 3,472 of those failures — roughly 27% of all company insolvencies in the year, from an industry that makes up a much smaller share of the economy's output (ASIC, FY2025–26; share approx.). Accommodation and food services accounted for roughly 15%, and "other services" — the ABS category covering personal services, repair and maintenance, mechanics, hairdressers and the like — for roughly 9% (approx., ASIC). Those three industries together carried a bit over half of all company failures in the year. All three share the same profile: small firms, thin margins, high labour intensity, discretionary or credit-financed demand, and heavy exposure to input costs they cannot hedge.
Why construction, specifically: four forces that compound
Construction did not fail because construction firms are worse at business than other firms. It failed because four separate pressures landed on the same balance sheets at the same time, and because the industry's contracting structure meant it could not pass any of them on.
One: fixed-price contracts signed before the shock, delivered after it. This is the mechanism at the centre of everything. A residential builder signs a contract in 2021 that fixes the price of a house. Construction then takes twelve to twenty-four months, and often longer once trade shortages bite. The builder buys the materials, and pays the trades, at 2022 and 2023 prices. Every input cost between signature and completion is the builder's to absorb. That is not an accident of any one company's contract drafting; it is the standard structure of Australian residential building, reinforced in some states by consumer-protection law that restricts price-escalation clauses on smaller domestic contracts. In an era of 2% inflation, the risk is trivial. In a shock, it is fatal, and it is fatal in a specific way: the more work a builder had on the books when the shock hit, the larger the loss. Growth — normally the marker of a healthy firm — became the thing that killed them.
Two: input-cost escalation. The ABS's producer price indexes for house construction record input and output costs rising sharply across 2021 to 2023, with the cost of building a house climbing on the order of a quarter to a third over that period (approx., ABS Producer Price Indexes, Australia — output of the house construction industry). Timber, structural steel, concrete, plasterboard and electrical components all repriced, some of them under genuine global scarcity. Labour repriced too: a shortage of trades in a market simultaneously running record detached-housing volumes, a large infrastructure pipeline and a public-sector building program bid up the price of every subcontractor on site.
Three: the cost of working capital. Construction is a sector that runs on other people's money. A builder finances materials and wages for weeks or months before a progress claim is certified and paid, and payment cycles in the industry are long. When the cash rate went from 0.10% to 4.35% between May 2022 and November 2023, the cost of every overdraft, equipment loan, and retention facility rose with it — and it rose for firms whose margins were already being compressed from the input side. This one is squarely the RBA's doing, exercised independently of government, and honesty requires saying so plainly: this was a monetary-policy consequence, not a Cabinet decision.
Four: the resumption of ATO debt collection. This is the most under-reported driver in the whole story, and the most directly governmental. Through the pandemic, thousands of construction companies deferred PAYG withholding, GST and superannuation guarantee liabilities, at the ATO's explicit encouragement. Total collectable debt across the tax system roughly doubled from its pre-pandemic level of around $26 billion at 30 June 2019 to above $50 billion by the middle of the decade, with small business owing roughly two-thirds of the total (approx., ATO annual report collectable-debt disclosures). Much of that was not a shortfall in profits owed to the Commonwealth. It was PAYG already withheld from employees' pay packets and GST already collected from customers, held on the balance sheet and used, in effect, as working capital.
When the ATO resumed firmer action, that arrangement ended over a compressed period. A company that had quietly been running on $400,000 of deferred tax found that money now had to be repaid on a schedule, that the interest on it was above 11% and compounding daily, that from July 2025 that interest was no longer deductible, that a director penalty notice could make the directors personally liable, and that the debt could be reported to credit bureaux — which, for a builder, threatens the trade credit and the surety bonding the business needs to keep operating at all. Several of those consequences arrive together, and the credit-reporting one is quietly the most dangerous: a construction company whose tax debt is visible to its suppliers can lose its trade terms in a week.
The Tax Office had, for two years, been the cheapest lender in the country. When that line of credit was called in, thousands of businesses discovered how much of their working capital it had quietly become.
None of this makes the ATO the villain. It was collecting money that was lawfully owed, that had been deferred rather than forgiven, and that in most cases had already been withheld from someone else. But the scale and the compression matter. A very large stock of debt, accumulated over two years of encouraged deferral, was called in over a much shorter window, and it was called in from exactly the industries — construction, hospitality, personal services — that were simultaneously absorbing an input-cost shock and a 425-basis-point rise in the cost of money.
The claim we will not make
It would be easy, and it would be wrong, to write that construction insolvencies are at a record and still climbing. They are not cleanly doing that. Construction failures in FY2023–24 sit at roughly 3,490 (approx., ASIC), which is marginally above the 3,472 recorded in FY2025–26. On the annual series, construction insolvencies have plateaued at a historically elevated level rather than continued to accelerate. The sharper trend claim that is properly supported is narrower and comes from a different cut of the data: in the financial year to March 2025, 2,636 construction companies became insolvent for the first time, up 23% year on year (ASIC, year to Mar-2025). First-time insolvencies are a useful measure because they exclude companies cycling through repeat appointments, and a 23% rise in first-time failures inside a plateauing total tells you the composition of failure was still worsening even as the headline count flattened.
The cleaner aggregate story is the total. Company insolvencies across all industries reached 14,152 in FY2025–26, and the December quarter of that year was the peak at 3,856 companies — the single worst quarter in the series (ASIC, FY2025–26). Being precise about which trend is and is not supported is not a hedge. It is the difference between an argument a well-informed opponent has to answer and one they can dismiss in a sentence.
What the number does not mean
Three pieces of context are needed before anyone concludes that Australian business is collapsing, because it is not.
First, the denominator. There are roughly 2.6 million actively trading businesses in Australia (approx., ABS Counts of Australian Businesses, including Entries and Exits, Jun-2025). A total of 14,152 company insolvencies is around half of one per cent of them. Construction, the worst-affected industry, contains well over 400,000 businesses (approx., ABS), so 3,472 failures is roughly three-quarters of one per cent of the sector. These are serious numbers, and they are concentrated numbers, but they are not a collapse.
Second, insolvency is not the same as business exit, and it is a small minority of exits. The ABS counts hundreds of thousands of business exits a year across the whole economy, the overwhelming majority of them voluntary — an owner retires, a sole trader takes a salaried job, a company is wound up solvently or absorbed into another (ABS Counts of Australian Businesses). Formal insolvency is the sharp, involuntary tail of a much larger and mostly ordinary process of churn.
Third, and most importantly for honesty: the aggregate business count has still grown across this period. Entries have exceeded exits, and the number of actively trading businesses in Australia is higher now than before the pandemic (ABS Counts of Australian Businesses, entries and exits series). Whatever else is true, this is not an economy in which businesses are net disappearing.
The RBA's own assessment has been consistent with that. Successive editions of the Financial Stability Review have noted company insolvencies rising back toward and past pre-pandemic levels while judging the failures to be concentrated in small firms with limited borrowings, and therefore a limited threat to the stability of the banking system (RBA, Financial Stability Review, 2023–2025). That is a fair reading of the financial-stability question. It is also not the question this topic is asking. A wave of small-firm failures can be simultaneously harmless to bank capital and highly damaging to the productive capacity of a specific industry the country urgently needs.
The critique: the problem is sequencing
The strongest criticism available here is not that any one of these measures was wrong. Taken on its own, each is defensible, and two of them were not decisions of this government at all.
The interest rate rises were the RBA's, made independently, and they were the correct response to an inflation rate that peaked at 7.8% (ABS CPI, Dec-2022). The tax debt was lawfully owed and had to be collected eventually; a permanent amnesty would have punished every business that paid on time. The fixed-price contract exposure was created by a global shock and by a contracting convention decades older than this government. The superannuation guarantee schedule was legislated years before this government took office.
The criticism is about what happens when all of them land on the same balance sheets inside the same eighteen months. Policy is not evaluated one measure at a time by the businesses it lands on. A builder does not experience "the resumption of ATO firmer action" and "the tightening cycle" and "the Closing Loopholes compliance changes" as three separate items in three separate portfolios. He experiences one cash-flow problem with four causes arriving at once, at the precise moment his contracted revenue is fixed and his costs are not.
Each measure was defensible on its own. The question no one owned was what they would do together, to the same firms, in the same window.
Three specific sequencing failures are worth naming.
The enforcement ramp had no industry-aware pacing. The ATO's return to firmer action was applied across the tax system on the basis of the debt and the taxpayer's engagement, which is administratively correct and legally proper. But the stock of deferred debt was not evenly distributed. It was heavily concentrated in exactly the industries carrying the input-cost shock. There is no public evidence of a mechanism by which the pace of debt recovery in construction was calibrated against what the RBA was simultaneously doing to the cost of working capital in the same industry. The Tax Office is not, and should not be, an economic-management agency. But that is precisely why the coordination had to come from somewhere else, and it is not clear that it came from anywhere.
The deductibility change landed at the worst point in the cycle. Removing the deductibility of the general interest charge from 1 July 2025 is a reasonable integrity measure in principle: it removes a tax subsidy for using the Commonwealth as a lender. But it was announced in December 2023 and took effect in July 2025 — that is, it was legislated into force during the period when the affected debt stock was at its largest and the affected industries were already failing at three times the suppressed-year rate. A measure designed to change future behaviour was applied to a stock of debt accumulated under an entirely different set of rules and explicit official encouragement. Retrospective in effect if not in law, its immediate incidence fell on businesses that deferred tax because the government of the day asked them to.
Nothing addressed the contract mechanism itself. The single most identifiable cause of construction failures in this period — fixed-price contracts exposed to an inflation shock, with escalation clauses restricted by state consumer law on smaller domestic contracts — attracted no significant Commonwealth or state legislative response while it was happening. There was no serious national process to examine whether domestic building contract risk should be shared differently in an inflation shock, no move to allow a limited, capped, independently-indexed escalation on contracts signed before a shock, and no federal-state work on how to protect homeowners without leaving the entire risk on the thinnest-capitalised party in the chain. The states retained the consumer protection, which is a legitimate policy choice. What neither level of government did was ask who else, if not the builder, should carry the risk of a once-in-forty-years input-cost spike.
There is a fair counter to be made on that last point, and it belongs here rather than in the steel-man: any escalation mechanism transfers price risk from builders to homebuyers, who are themselves at the limit of their borrowing capacity and are the party the law was written to protect. That is a genuine trade-off with no costless resolution. But "hard trade-off" is a reason to have the argument in public, not a reason to leave the default in place by inaction while three thousand construction companies a year fail.
The honest verdict, on the site's own standard, is a division of responsibility. The government did not cause the inflation shock, the global supply-chain disruption, the fixed-price contracting convention, or the rate rises. It inherited a suppressed insolvency backlog, a $50-billion-plus stock of deferred tax debt, and a superannuation schedule already in law. What it presided over, and what it owned, was the timing: the decision about how fast to unwind the deferral, whether to phase the deductibility change, and whether to sequence the industrial-relations compliance changes against what was happening in the worst-affected sectors. Timing was the variable actually within the government's control, and timing is where the case against it sits.
What they got right, and the honest limits
Three things complicate this story, and they need saying plainly rather than buried.
The pandemic support genuinely worked. JobKeeper, the insolvent-trading moratorium, the elevated statutory demand threshold and the ATO's forbearance kept a very large number of viable businesses alive through a period when their revenue had been shut off by government order. The evidence is in the numbers: insolvencies ran at roughly half their normal rate for two years during the sharpest economic contraction since the 1930s (approx., ASIC Series 1A). Businesses that would have failed did not. Many of them are still trading. That was expensive, it was blunt, and the ANAO and others have documented real design flaws in JobKeeper — but the central objective was achieved, and it was achieved by a Coalition government whose work on this the current government did not undo.
The ATO did not simply switch enforcement back on uniformly. Its published approach through the resumption prioritised engagement over enforcement: businesses that made contact and entered a payment plan were, as a matter of stated policy, treated differently from those that did not, and the ATO has consistently reported large numbers of active payment arrangements alongside its firmer action (ATO annual reports and published debt-collection guidance). Firmer action was aimed disproportionately at non-engagement — at the businesses that ignored letters — rather than applied as a blanket collection sweep. That distinction is real, it was deliberate, and the criticism above is about the pace and the coincidence of the ramp, not about a claim that the ATO acted indiscriminately.
And the business population has grown. Across the whole period examined here, business entries exceeded business exits, and Australia has more actively trading businesses now than before the pandemic (ABS Counts of Australian Businesses, entries and exits). Any account of this period that leaves a reader believing the business base is shrinking has misled them.
One further credit is due on the insolvency system itself. The Small Business Restructuring regime, created by the Corporations Amendment (Corporate Insolvency Reforms) Act 2020 and available from 1 January 2021, lets an eligible company with liabilities under $1 million restructure its debts while the directors stay in control, at a fraction of the cost of a full voluntary administration. Appointments under it have grown strongly through this period. Some part of the rise in formal appointments in the ASIC series reflects distressed small companies using a cheaper mechanism that simply did not exist before 2021 — a genuine improvement in the plumbing, and a small but real caveat on the raw insolvency count.
The honest limit on the steel-man is this. "It is mostly catch-up" is a strong argument for 2023 and defensible for 2024. It gets weaker each year it is repeated. A backlog is by definition finite: the businesses that would have failed in 2020 and 2021 can only fail once. By FY2025–26, four to five years after the moratoria lapsed and five years after JobKeeper ended, a total of 14,152 — running well above the pre-pandemic rate rather than converging back toward it, and peaking in the December quarter of that same year — is harder to explain as deferred 2020 failures. At some point the catch-up explanation has to give way to an explanation about current conditions. That point has arrived, and the December-quarter peak of 3,856 is the evidence for it.
The risk to the future
Losing the builders the housing target depends on
Australia has committed to 1.2 million new well-located homes over the five years from July 2024 under the National Housing Accord — roughly 240,000 dwellings a year, against completions that have been running closer to 170,000 to 180,000 (approx., ABS Building Activity, Australia). Delivering that requires more building capacity, not less. What the insolvency data records is the steady removal of small and mid-sized builders and trade contractors — 3,472 construction companies in a single year, and a 23% rise in first-time construction insolvencies in the year to March 2025 (ASIC). Capacity does not transfer cleanly when those firms fail. The working capital is gone, the equipment is sold at auction, the licensed supervisor who could run three sites at once takes a salaried job or leaves the industry, and the apprentices go with them. Directors who have been through a personal-liability event are slow to start again, and banks and sureties are slow to back them. There is a second-order risk in what replaces them: as small and mid-sized builders exit, market share concentrates in the largest firms and in the volume operators with the balance sheets to absorb a shock — which reduces competitive tension on price in exactly the market where affordability is already the binding constraint. The forward risk is not a financial crisis; the RBA is right that small-firm failures do not threaten bank capital. The risk is slower and harder to reverse: an industry that emerges from this period smaller, more concentrated, more risk-averse about fixed-price work and short of the skilled supervision that takes a decade to rebuild, at exactly the moment the country has staked its housing policy on that industry building more than it ever has.
The builder in the growth corridor is not coming back. His nine houses were completed and are lived in; the families in them got the price they were promised, which was the point of the law that made him carry the risk. His concreter took a wage job with a larger firm. His two apprentices did not finish. The trade credit, the site supervision and the twenty-year relationships with suppliers that made his business work are dispersed and will not reassemble. Some of that is the ordinary, necessary churn of a market economy clearing a bad bet, and it would be dishonest to pretend otherwise. But the bet he lost was made with 2021 information about 2023 prices, and the reckoning arrived alongside a tax bill deferred at the government's own invitation, a cost of money that had risen more than four percentage points, and a compliance regime being rewritten in the same window. No single one of those was aimed at him. All of them arrived together, and he is not the only one who did not survive the arrival.