The slab was poured in March and by August it was still bare. On a growth-corridor estate at the edge of an Australian capital — the kind with a display village at the entrance and unsealed roads at the back — the site supervisor walked the same twelve lots most mornings and counted the same problem. Four were waiting on a bricklaying crew. Two were waiting on a plumber who had been pulled onto a hospital job that paid better and ran longer. One was waiting on a truss delivery rescheduled twice. None of the twelve was waiting on a buyer, a bank, or a council. Every one of them had a customer with finance approved and a permit already issued. They were waiting on people, and there were not enough of them.
Two of those lots belonged to a couple in their early thirties who had signed a fixed-price contract eighteen months earlier and had been renting ever since, paying a mortgage-sized rent on one property while paying interest on a construction loan for another. Their builder had not collapsed and their contract had not been repudiated. They were simply in a queue that was not moving at the speed the paperwork implied. The supervisor and the couple are composites — invented, but assembled from the pattern the national data describes. That pattern is the subject of this article: Australia has approved and contracted for far more homes than it is physically completing, and the gap between the two is now the central fact of national housing policy.
What the government did: a number, and the machinery attached to it
On 16 August 2023, National Cabinet — the Prime Minister, every premier and chief minister, and the president of the Australian Local Government Association — agreed the National Housing Accord. Its headline commitment was 1.2 million new, well-located homes over the five years from 1 July 2024 to 30 June 2029 (National Cabinet, National Housing Accord communiqué, Aug-2023). The figure was an upgrade on an earlier aspiration of one million homes floated in the October 2022 Budget, when the Accord was first conceived as an agreement between the Commonwealth, the states, institutional investors and the construction industry.
The words "well-located" were doing real work. The Accord was not a request for more dwellings anywhere. It was a request for dwellings near jobs, transport and services — in practice, a request that state and local planning systems permit more density in established suburbs rather than only more sprawl at the fringe. That is the hardest thing in Australian housing policy to do, and the Accord's principal achievement is that it made states attempt it.
Around the target the Commonwealth built a set of instruments. A $3 billion New Home Bonus pays states and territories for homes delivered above an agreed baseline, at a rate in the order of $15,000 per home (approx.; Treasury). A $500 million Housing Support Program funds planning capacity and enabling work at the state and local level (approx.; Treasury). The National Housing Infrastructure Facility — concessional finance for the water, sewer, power and road connections a new estate needs before a single frame goes up — was expanded (approx.; Housing Australia). And the National Housing Supply and Affordability Council (NHSAC) was established as an independent statutory advisory body, publishing an annual State of the Housing System report and quarterly monitoring against the Accord.
It is worth being exact about what the Commonwealth can and cannot do here, because most public argument about the target gets this wrong in one direction or the other. The Commonwealth does not build houses. It does not issue planning permits, zone land, or employ carpenters. Australian dwellings are overwhelmingly built by private firms, financed by private lenders, on land whose permitted use is set by state planning instruments and local councils. What the Commonwealth controls is money, tax settings, migration, and the convening power to get eight other governments into a room. The Accord is an attempt to use the last of those to move the rest.
That does not make the target meaningless — the Commonwealth chose the number, published it and attached payments to it — but responsibility for the outcome is genuinely shared. This article separates what the current government caused, what it inherited, and what it merely presided over. The productivity problem is a thirty-year, multi-government, international phenomenon. The interest rate cycle that broke development feasibility from 2022 was monetary policy set by an independent central bank. State planning systems are not Commonwealth policy. What the Commonwealth owns is the target, the incentives attached to it, and the decision to set a number without a matching plan for the capacity to deliver it.
The strongest case for the policy
A stretch target is a legitimate instrument, and this one moved things nothing else had moved
Put the government's case at its best and it is considerably stronger than the raw shortfall numbers suggest. Start with the nature of the target. A stretch target is a recognised and legitimate tool of government, used as it is used in corporate strategy and public health: set above the comfortable expectation to mobilise effort, reorder priorities and force institutions to justify why they cannot do better. Judged that way, the question is not "was the target hit" but "did more homes get built than would have been built without it". A target missed by a wide margin can still be a policy success if it shifted the trajectory upward.
Second, the Accord did something no Commonwealth housing initiative had achieved in decades: it made state planning reform a matter of national accountability with money attached. The New Home Bonus ties payments to state performance against a share of the national target, converting an abstract federal aspiration into a state treasury line item. States have since moved on things they had avoided for years — low- and mid-rise housing reforms and transport-oriented rezoning in New South Wales, activity-centre planning and standardised townhouse codes in Victoria, comparable measures elsewhere. Planning reform is slow, locally unpopular work with no ribbon to cut. It does not happen without external pressure, and the Accord supplied it.
Third, the causes of the shortfall lie overwhelmingly outside Commonwealth control. Between May 2022 and November 2023 the Reserve Bank of Australia raised the cash rate target from 0.10 per cent to 4.35 per cent — the fastest tightening cycle in a generation (Reserve Bank of Australia, cash rate decisions). That did two things to supply at once: it raised the cost of development finance, and it cut the borrowing capacity of the buyers whose pre-sales developers need to secure that finance. Projects that stacked up in 2021 did not stack up in 2023. This was monetary policy conducted independently to bring down inflation, and no federal housing minister of any party could have prevented it.
Fourth, the productivity problem the second half of this article documents is not this government's creation, or any recent government's. It has been building for three decades, it appears in comparable form in the United Kingdom, Canada and the United States, and it is the subject of a substantial international research literature. Blaming a government elected in 2022 for a trend that began before the internet is not analysis.
Finally, the alternative to an ambitious target was not a modest, achievable one. It was the status quo ante, in which no Australian government had a published national dwelling target at all, no state faced a national benchmark, and housing supply competed for attention with everything else. The Accord made supply the organising question of national housing debate. That is a real achievement and it survives the shortfall.
The evidence: the arithmetic, which is not in dispute
The virtue of this topic is that its central claim requires no interpretation, no modelling and no contested assumption. It requires division.
The target is 1.2 million homes over five years. Divided evenly, that is 240,000 completions a year, or 60,000 a quarter. There is nothing arbitrary about that arithmetic; it is the number the Accord itself implies, it is the number Treasury and NHSAC use to monitor progress, and it is the number every state's share is calculated from.
In the year to March 2026, Australia completed 173,400 dwellings (ABS Building Activity, Australia, Mar-2026). That is 66,600 homes below the required annual run-rate — a shortfall of about 28 per cent. Put in human terms, the required pace is one completed home every two minutes and eleven seconds, around the clock, every day of the year. The actual pace was one every three minutes.
The cumulative position is worse, because the shortfall has been running since the Accord's clock started. Over the first 21 months — July 2024 to March 2026 — Australia built 112,400 fewer homes than the target required, a cumulative shortfall of about 27 per cent (NHSAC monitoring; ABS Building Activity). Against a required 420,000 homes for that period, actual completions were therefore in the order of 307,600 (derived from the published shortfall). At average Australian household size, the missing 112,400 dwellings are housing for roughly 280,000 people (approx.; ABS Census household size) — close to the entire dwelling stock of a mid-sized regional city, absent from the market.
What catching up would now require
This is the calculation that settles the question, and it is worth showing the working so that any reader can check it.
If completions to March 2026 stand at roughly 307,600, then 892,400 homes remain to be built by 30 June 2029. That leaves 13 quarters. Dividing one by the other gives a required rate of approximately 69,000 completions per quarter for the entire remainder of the Accord period — about 275,000 a year (author's calculation, derived from the published target and the published shortfall; treat as derived, not as an official figure).
Set that against the record. Recent quarterly completions have been running at about 43,350 (173,400 divided by four). The catch-up rate is therefore roughly 59 per cent above the pace the industry is currently achieving, sustained for more than three years without a single weak quarter. It is also above the highest annual figure Australia has ever recorded: the peak of the apartment boom in the middle of the last decade produced something in the order of 220,000 completions in a year (approx.; ABS Building Activity, 2016-17), which remains the national record. The Accord's original 240,000 was already above that record. The catch-up rate is roughly a quarter above it again.
Australia has not asked its construction industry to do something hard. It has asked it to exceed its own all-time record every year for three years, using a workforce that builds fewer homes per hour worked than it did in the 1990s.
Australia's three leading private-sector housing forecasters have separately concluded that the 1.2 million target will not be met (secondary sources; independent forecasting houses, 2025-2026 publications). That is not a primary record, and forecasters are often wrong. It is cited here only because unanimity among competitors who normally differ is itself informative.
One qualification belongs on the record. The target has not formally failed. Three financial years remain, and completions are a lagging indicator: a dwelling completed in 2028 was approved in 2026 and financed in 2025. That cuts both ways — recent reform has not yet had time to appear in the data, and the homes needed to close the gap by 2029 would already have to be in the pipeline. This article therefore argues about trajectory and probability, not about a settled outcome.
Why the homes are not being built
The shortfall is the easy half of the analysis. The explanation is the interesting half, and it leads somewhere more uncomfortable than an argument about targets. Australia is not failing to build homes for want of demand, land, approvals or capital; on several of those measures it has more than it had a decade ago. It is failing because the industry that builds them has been getting less efficient for thirty years.
The productivity problem, stated carefully
Labour productivity is simply output per hour worked. Across the Australian market sector as a whole it has risen substantially since the mid-1990s — slower recently than historically, but the direction has been up. Construction is the conspicuous exception. On the measure that matters most here — new dwellings completed per hour worked in residential construction — the Productivity Commission has found a decline of roughly half since the mid-1990s (approx.; Productivity Commission research on housing construction productivity, 2025). On broader value-added-per-hour measures the decline is smaller and more contested, because construction output is hard to measure when the product itself changes: today's dwelling is larger, better insulated, more fire-resistant, more accessible and more energy-efficient than a 1995 dwelling, and quality gains that do not show up as extra units of output read as falling productivity.
That caveat is real and should temper the headline. It does not dissolve it. Even allowing generously for quality improvement, the sector has not delivered the compounding efficiency gains that manufacturing, retail, logistics, mining and telecommunications have delivered over the same period. An industry that builds fewer houses per hour than it did a generation ago cannot expand output rapidly by being asked to. It can only expand by adding hours — that is, by adding workers — and the workers are the binding constraint.
Fragmentation: an industry of very small firms
Australian residential construction is among the most fragmented industries in the economy. Construction consistently records more actively trading businesses than any other industry division in the ABS business counts, and the overwhelming majority are non-employing sole traders or micro-businesses (ABS, Counts of Australian Businesses). A typical detached house is built not by a company but by a builder coordinating twenty to thirty independent subcontractors, most of them one-to-three-person operations.
That structure has genuine advantages: it is flexible, it responds fast to local conditions, and it keeps overheads low. But it is close to the worst possible structure for productivity growth. A two-person subcontracting business cannot fund research and development, cannot amortise expensive plant across enough jobs to justify buying it, and has little capacity to train apprentices properly while also meeting this week's invoices. Productivity gains in other industries have come overwhelmingly from firms large enough to invest in capital, systems and process. Residential construction has very few such firms, and those it has are concentrated in high-rise rather than the detached and low-rise segments where most of the Accord's homes must come from.
Low capital intensity and slow technology adoption
The consequence is a sector unusually labour-intensive and unusually resistant to the technologies that transformed comparable industries. Most Australian homes are still built the way they were built in 1990: components delivered to an open site, assembled outdoors by successive trades, subject to weather, sequencing failures and rework. Prefabricated and modular construction — where the building or its major components are made in a factory and assembled on site — remains a small share of Australian residential output (approx.; industry and Productivity Commission analysis), far below its share in several comparable economies.
The reasons are structural rather than cultural. Factory-built housing requires steady, high-volume order flow to justify the capital investment, and Australian residential demand is famously cyclical, with peak-to-trough swings that have bankrupted more than one prefabrication venture. It also runs into planning and building-approval systems that differ between states and, in places, between councils, fragmenting the national market a factory would need. And it runs into finance: lenders' progress-payment structures are built around on-site construction stages, and do not always accommodate a build where most of the value is created before anything appears on the block.
A target is a number. A house is a trade, a slab, a supply chain, and a bank that will lend against it.
The collapse of fixed-price contracting
The most acute recent shock to delivery capacity was neither productivity nor planning. It was the failure of the risk-allocation model the industry had used for decades.
Australian residential building runs on fixed-price contracts. The builder quotes a price, the customer arranges finance against that price, and the builder absorbs any movement in input costs between signing and completion. That works when input prices are stable. From 2021, they were not. Construction input costs rose sharply as global supply chains seized, timber and steel prices spiked, and domestic demand was inflated by stimulus that had pulled a large volume of detached-house commencements into a narrow window. Producer price measures of construction inputs and outputs recorded increases far above general inflation across 2021 and 2022 (ABS, Producer Price Indexes, Australia).
Builders holding contracts signed in 2020 and 2021 were therefore obliged to complete homes in 2022 and 2023 at prices that no longer covered their costs. Many did not survive it. Construction has for several years accounted for the largest share of company insolvencies of any industry in Australia (ASIC, insolvency statistics). Each failure is a small catastrophe for the customers and subcontractors involved, and it destroys delivery capacity that does not repair quickly: project management knowledge, subcontractor relationships, supplier terms and trained staff all disperse.
The second-round effect matters more for the Accord than the failures themselves. Surviving builders responded rationally: shorter price validity periods, rise-and-fall clauses where they could get them, larger contingencies, and far more selectivity about which projects to take on. That is prudent firm behaviour and it is also, in aggregate, a reduction in the industry's willingness to commit capacity at any given price. An industry that has just been burned does not sprint into a stretch target.
Planning and approval friction — a state and local constraint
Planning is where the Commonwealth's writ does not run, and where a great deal of the delay lives. Assessment timeframes vary enormously between states and between councils within states, and multi-dwelling proposals in established areas routinely face appeals measured in months or years. Each month of delay is a month of holding costs on land, and holding costs are what tip a marginal project from feasible to abandoned.
Two points of fairness apply. First, this is precisely the constraint the Accord was designed to attack, and it is where the Accord has produced the most genuine movement. Second, planning delay is not simply obstruction: the systems exist to manage flood risk, bushfire exposure, heritage, infrastructure sequencing and neighbourhood amenity, and communities have legitimate interests a purely volume-maximising system would run over. The critique is not that planning assessment exists. It is that these systems were designed for lower throughput and have not been re-engineered for the volumes now demanded of them.
The trades shortage
Every constraint above resolves into the same one: people. Australia's national skills assessments have persistently identified core construction trades — bricklayers, carpenters and joiners, plumbers, electricians, plasterers, roof tilers, concreters — as in shortage across most or all jurisdictions (Jobs and Skills Australia, Occupation Shortage List / Skills Priority List, recent editions). This has compounding sources rather than a single cause.
Apprenticeship completion is the first. Construction trade apprenticeships have long had completion rates in the vicinity of half of commencements (approx.; NCVER apprentice and trainee statistics), which means the training system must recruit roughly two apprentices for every tradesperson it ultimately produces, with a three-to-four-year lag before that person is productive and longer before they can supervise. A policy response to a trades shortage therefore cannot deliver inside the Accord's own timeframe: a decision taken in 2024 to double apprentice intake produces qualified tradespeople from about 2028.
Demographics are the second. The trade workforce skews older than the workforce average in several occupations, and retirements are removing experienced supervisors — the people who make sites run efficiently — faster than training replaces them. The third is the attractiveness of the work relative to alternatives, in an economy where care, logistics and services have absorbed large volumes of labour.
The infrastructure pipeline: competing for the same people
The final constraint is least often stated plainly, because it involves governments competing with their own housing objectives. Australian governments, state and federal, have been running the largest public infrastructure program in the country's history — metro rail, road tunnels, level crossing removals, hospitals, energy transmission and, increasingly, defence facilities. Those projects need concreters, electricians, plant operators, surveyors, steel and cement. They draw from the same pool as housing.
They also outbid it. A major public project offers longer engagements, higher rates and more certain payment than a residential builder can match, and the differential is large enough to move labour at scale. Infrastructure Australia's market capacity assessments have documented this competition explicitly, identifying shortfalls in the workers and materials needed to deliver the announced public pipeline (Infrastructure Australia, Infrastructure Market Capacity Report, recent editions). In 2023 the Commonwealth conducted an independent strategic review of its own Infrastructure Investment Program and then removed or reprofiled a substantial number of projects — in the order of 50 (approx.; Department of Infrastructure, Transport, Regional Development, Communications and the Arts, 2023) — with construction market capacity among the reasons cited. That was an implicit official acknowledgement that the public pipeline and the housing target draw on one finite resource.
This is a genuine trade-off rather than an error, and it deserves to be described as one. The hospitals and rail lines are also needed, and several state programs were committed before the Accord existed. But the trade-off was never made explicitly, and the housing target was set as though the workforce constraint did not bind.
Approvals are not becoming homes
All of these constraints converge on a single, measurable symptom: the widening distance between a dwelling being approved and a dwelling being finished.
Building approvals and building completions are two different ABS collections measuring two different events. An approval is a permit. A completion is a house. Historically the two series tracked each other with a reasonably stable lag, completions following approvals a year or two later at a fairly predictable conversion rate. That relationship has loosened. Approvals in recent periods have not converted into completions at historical rates, and the average time from approval to completion has lengthened (ABS, Building Approvals, Australia; ABS, Building Activity, Australia).
Part of that lengthening is compositional and entirely benign: apartments take far longer to build than detached houses, and as the dwelling mix shifts toward higher density near transport — exactly what "well-located" requires — the average approval-to-completion time mechanically rises.
The rest is not benign. Approvals lapse. Approved apartment projects sit uncommenced because the developer cannot achieve enough pre-sales at prices covering construction cost, finance cost and a margin that justifies the risk. Approved detached lots sit unbuilt because the builder cannot secure a slab crew, or the customer's finance approval expired while they waited. A permit exists, a site exists, demand exists, but the home does not. That is the failure mode distinguishing this shortfall from previous downturns: not a demand failure or an approvals failure, but a delivery failure.
The critique: a target set without a plan for the constraint that binds
The criticism that follows is narrower and more serious than "the government missed a target". Governments miss targets constantly. The problem here is a mismatch between the instrument chosen and the constraint that actually determines the outcome.
The Accord's machinery — the New Home Bonus, the Housing Support Program, the infrastructure facility, the planning commitments — is almost entirely aimed at the approvals and enabling-works end of the process. That was defensible: planning was a real constraint, and the one the Commonwealth could plausibly influence. But the evidence above indicates the binding constraint is now further down the chain. Adding incentive payments for approvals into a system that cannot convert the approvals it already has is pushing on a rope.
Nothing in the Accord's original design directly addressed labour supply, apprenticeship completion, prefabrication capacity, the fixed-price contracting risk model, or the competition between the housing target and the public infrastructure pipeline. Measures on several of those fronts have followed. But they followed the target rather than accompanying it, and their lags run beyond 2029. A target with a five-year clock was attached to constraints with a five-to-ten-year response time.
The second criticism concerns the number itself. Ambition is legitimate. But there is a difference between a stretch target set 10 or 15 per cent above the best previously achieved and one set above the all-time record from a starting point 28 per cent below it. Once a target passes the threshold where practitioners privately regard it as arithmetically impossible, it stops mobilising effort and starts doing the opposite: it becomes a number to be managed, explained and eventually quietly re-based. Credibility is a finite resource for governments, and targets spend it.
The third criticism is about measurement, and it receives the least attention. The Accord counts homes. It does not, in any published national series, measure how many of them are "well-located" — the adjective that carried the entire policy rationale for density near jobs and transport. A target whose qualifying adjective is unmeasured will, under pressure, be met by whatever is easiest to count. The path of least resistance in Australian housing has always been fringe subdivision, which is cheaper and faster to approve, and which puts homes precisely where the Accord's own language said they should not principally go. Meeting the number while failing the adjective would be worse than the current shortfall, and nothing published allows the public to tell whether it is happening.
The fourth concerns accountability design. The Accord distributes credit and blame in a way that makes neither stick. The Commonwealth set the target but does not control the levers. The states control the levers but did not set the target and face an incentive payment that is modest against their own budgets. The private sector does the building but signed nothing. When the target is missed, each participant will have a defensible account of why it was not their doing, and each account will be substantially true. An arrangement in which everyone is responsible and no one is accountable will not correct itself.
What they got right, and the honest caveats
Three things belong on the government's side of the ledger, stated plainly, because the critique above is only credible if the credit comes first.
The Accord genuinely moved state planning policy. This is the most important and least appreciated part. For thirty years, Commonwealth housing initiatives ran into the wall of state and local planning autonomy and stopped there. The Accord, by attaching money and a published national benchmark to state performance, produced movement on questions state governments had actively avoided: permitting mid-rise housing in established suburbs, rezoning land around transport nodes, standardising codes so a compliant townhouse does not require a bespoke assessment in every municipality. These reforms are politically costly locally, deliver no ribbon-cutting, and generate organised opposition in exactly the electorates governments least want to antagonise. Doing them anyway is hard, unglamorous public policy.
The target kept supply at the centre of national debate. Before 2023 the Australian housing conversation was overwhelmingly about demand-side measures — grants, guarantees, deposit schemes, tax settings. The Accord reframed it around whether homes exist. That reframing is analytically correct and has outlasted the specific programs attached to it.
Dwelling approvals have improved in some jurisdictions, and the machinery built alongside the target has value beyond it. The National Housing Supply and Affordability Council is a permanent, independent statutory body publishing rigorous analysis of the housing system, including analysis unflattering to the government that created it — the very reporting this article relies on to quantify the shortfall. Creating an institution capable of publicly documenting your own policy's underperformance is a durable improvement that outlives a five-year target.
The caveats on the credit are equally honest. Planning reform is necessary but not sufficient, and the evidence here suggests it is no longer the binding constraint. Approvals improving while completions lag is exactly the pattern of reform working at one stage of the pipeline and failing at the next. The labour and productivity measures that would address the actual constraint arrived later and cannot deliver inside the target window. And a shortfall of 112,400 homes in 21 months is not a rounding error in an otherwise successful mobilisation — it is a signal that the mobilisation did not reach the part of the system that matters.
One further piece of fairness. This policy may well look better in retrospect than the 2029 scoreboard suggests. If the planning reforms extracted through the Accord raise Australia's sustainable building capacity through the 2030s, they will produce far more than 1.2 million homes over time, and the target that failed will have been the price of the reform that worked. That is a real possibility and this article does not exclude it. But it is a hope about a later decade, and the people in the vignette at the top of this page are waiting now.
The risk to the future
The supply response does not arrive, and the shortfall compounds into another cohort
The forward risk is not that a number in a communiqué goes unmet; targets are missed all the time and the sky stays where it is. The risk is that the 1.2 million figure was the country's answer to a genuine emergency — the answer given to renters facing the sharpest sustained rent increases in decades and to first home buyers watching the deposit hurdle rise faster than they can save — and the answer does not arrive. Every home not built is a household that stays in the rental market, competing with the households already there. Shortfalls in housing do not clear; they queue.
The compounding is what makes it serious. A deficit of 112,400 homes in the Accord's first 21 months does not sit still while the industry catches up, because underlying demand keeps accruing from household formation, migration and the long-run trend to smaller households. If completions stay near current rates, the gap at June 2029 will be several times its present size, inherited whole by whoever is governing then. Homes not built in 2025 and 2026 are not deferred to 2030. They are subtracted permanently, and the people who would have lived in them spend those years somewhere more expensive, more crowded, or further from work.
The generational shape of that risk is the part worth sitting with. Australians now in their late twenties and early thirties are the cohort the Accord was implicitly written for. If the supply response lands in the 2030s instead of the 2020s, it arrives after that cohort's household formation decisions have already been made under scarcity — later partnering, later children, more years of rent paid rather than principal repaid, less lifetime wealth in the asset the Australian retirement system quietly assumes people will own. The affordability problem does not simply persist. It hardens around a specific group of people, as it hardened around the cohort before them.
None of this is a prediction that the target will fail. Three years remain, reforms already legislated in several states have not yet shown up in completion data, interest rates have begun easing from their peak, and construction cycles can turn faster than they appear to from inside a trough. Those possibilities are genuine and a fair reading holds them open. But the probabilities are not symmetrical, and the arithmetic is not ambiguous. Reaching 1.2 million now requires the industry to sustain, for thirteen consecutive quarters, an output rate it has never achieved once. The honest statement of the risk is not that this will not happen. It is that nothing in the current evidence explains how it could.