It Only Works If It Goes Up

Australia runs its housing system at nine times the volume it was built for, then calls the result a market failure.

On 25 August 2026, a Sydney developer with more than 520 subsidiaries and roughly 15,000 dwellings in its pipeline handed itself to administrators and described the circumstances as a perfect storm.

There was no storm. There was arithmetic, arriving on schedule.

Bathla Group’s principal entity, Universal Property Group, reported $3.2 billion in liabilities as at June 2025, with total group financing exposure reported between $3.5 and $3.6 billion, most of it private credit rather than banks. Alceon exited roughly $670 million in January 2026. By July, financiers had taken effective control of parts of the pipeline and were paying subcontractors directly to keep sites moving. The managing director cited softening sales, the May budget changes, and falling confidence. About 2,000 of those dwellings were actually under construction. In New South Wales alone, 1,522 construction firms collapsed in 2025-26.

A business that can’t survive prices not rising was never a housing business. It was a position on a price, wearing a hard hat.

That distinction has been lost in this country for 50 years, and it’s been lost deliberately, because an enormous amount of capital, revenue, and political survival now depends on nobody making it. Prices are falling. The correction is the first honest signal the Australian housing market has produced in my working lifetime. Every institution with a claim on the old number is working to have it stopped, and whether they succeed will decide what a house costs here for the next generation.

What a house is for

Prices should keep falling. They should keep falling until one income buys a house, and then they should sit inside a narrow band and move very little, because that’s what a price does when it’s measuring shelter rather than carrying a bet.

That standard was the arrangement in this country within living memory, not a piece of nostalgia. One earner, one household, one mortgage, and the other adult free to do the work that actually holds a family together, which is work whether or not anyone pays a wage for it. Somewhere between then and now the second income stopped being a choice and became a requirement, and here’s the part nobody says out loud. The household gained nothing by it. It roughly doubled its labour input into the market and bought the same house. The second wage was absorbed by the price.

And it didn’t stop there. Every affordability measure the industry publishes now quietly concedes the argument, because every one of them is calculated against household income rather than a wage. Servicing a new mortgage on the median dwelling takes 45.9 per cent of median household income, against a long-run average around 34 per cent. Saving the deposit takes 11.2 years, a record, while 33 per cent of household income goes on rent in the meantime, also a record. Two incomes, nearly half the combined gross before tax or food or childcare, and 11 years of saving on whatever’s left. The share of median-income households able to afford a median-priced home has fallen to roughly 14 per cent, down from 43 per cent. In February 2026, weeks before the peak, the national median dwelling was $922,838.

So the sequence isn’t one income to two. It’s one income, then two, then neither.

Look at what the number did to get there. The Sydney median house price sat at roughly five times average annual earnings in 1970. By 2024 it was above 20 times. Nothing about the physical stock changed to justify that. A three-bedroom house on a suburban block didn’t quadruple in usefulness. What changed was the function the asset was asked to perform. At some point Australia stopped treating housing as shelter and started treating it as an industry, with a lobby, a media vertical, a finance layer, a tax scaffold, and a national mythology attached that we still call the Australian dream while charging admission to it.

Now look at where the number is. NAB has revised its 2026 forecast to declines near 10 per cent in Sydney, 9 per cent in Melbourne, and 2 to 4 per cent across the mid-sized capitals, with the year’s falls not yet finished. Cotality recorded a national fall of 0.7 per cent in July 2026, the sharpest monthly decline since December 2022. Over the quarter, the upper quartile fell 3.2 per cent. NAB’s chief economist puts the downturn at roughly a third of the way through.

A third of the way through isn’t a crisis. It’s a start.

Who is hurt by a house getting cheaper? Investors carrying leverage they only ever intended to service out of appreciation. Developers whose model requires the settlement price to exceed the feasibility price by more than the cost overrun. Lenders holding books written against valuations that assumed a permanent direction of travel. And the Commonwealth, whose forward estimates now have a specific and datable interest in the recovery.

Who benefits? People who want somewhere to live.

Who gets asked

A recent segment on one of the mainstream networks laid the apparatus out neatly enough that it deserves to be studied rather than argued with.

The presenter opens by observing that the Prime Minister is happy with the effect the tax changes are having on the property market, and that first home buyers are facing less competition from investors. Note that this is presented as the accusation. The guest, a real estate auctioneer, then advances a theory. Treasury would’ve understood, he suggests, that lower valuations before 1 July 2027 mean a higher capital gains take once prices recover, because that date sets the baseline. He says there’s no clear evidence for it. He says it might be a conspiracy. He says both of those things before anyone else in the studio has the chance to.

That’s the move. Supply the discrediting word yourself, up front, and the mechanism underneath it never has to be examined by anybody. The idea gets filed as speculation instead of tested as design, and the audience leaves with the impression that only cranks look at incentives.

Then consider who gets to be the expert witness across that segment. An auctioneer and a property developer. Two people whose income is a function of transaction volume and price. Nobody appears who rents. Nobody appears who’s been priced out. And the thing being reported as a catastrophe, in the framing and the tone and the choice of guests, is first home buyers winning auctions.

Not a conspiracy. A calendar.

The mechanism is real. It’s also public, legislated, and considerably larger than the segment let on.

From 1 July 2027, the 50 per cent capital gains tax discount disappears. It’s replaced by cost base indexation plus a minimum 30 per cent tax on net capital gains for assets held longer than 12 months. This isn’t a property measure. It applies across all capital gains tax assets held by individuals, trusts, and partnerships, including pre-1985 assets that have been exempt for four decades. It’s law, carried by the Treasury Laws Amendment (Tax Reform No. 1) Act 2026. Negative gearing on established residential property acquired after 7.30pm on 12 May 2026 goes at the same time, with losses quarantined against rental income and carried forward rather than deducted against wages. New builds keep both concessions. Holdings that existed before budget night are grandfathered.

The auctioneer’s version of the baseline argument doesn’t survive contact with the legislation, and that’s worth correcting rather than repeating. Nobody’s compelled to take a 1 July 2027 valuation. A taxpayer can elect a time apportionment formula instead, splitting the gain by the period held under each regime. Run the argument in the form he ran it and any competent adviser dismantles it in a sentence, which is how a real point gets buried.

Here’s the version that holds.

Indexation strips out inflation. It does not strip out recovery. An owner whose property falls 20 per cent and then climbs back to exactly where it started has made nothing. In real terms they’re square. Under the new regime, that climb is a capital gain, taxed at a floor of 30 per cent, because the code measures from the bottom of the hole rather than from the edge of it. The deeper the fall before the baseline sets, the larger the taxable recovery afterwards.

So the question isn’t whether anyone in Treasury wanted valuations soft before a particular date. The question is what happens after it. From 1 July 2027, the Commonwealth holds a direct, quantified, budgeted claim on Australian house prices going back up. That claim now sits inside the forward estimates. Every officer whose job is to defend those estimates has an interest in the number recovering, and none whatsoever in it settling.

That’s the thing to refuse. Not the tax change. The assumption underneath it.

The same segment produced the standard rebuttal, delivered by a developer: these reforms will drive rents higher than the 1980s. That comparison fails on its own terms. The 1985 to 1987 episode was an outright quarantine with no grandfathering and no carve-out for new supply. This is neither. And the record doesn’t say what it’s used to say. Nominal rents rose nationally through a high-inflation period, but real rent growth was confined to Sydney and Perth, both already sitting near 1 per cent vacancy before the change, while real rents fell in Adelaide and Brisbane. The Hawke government’s own 1987 cabinet papers attributed the increases to local market conditions rather than the tax measure. A landlords’ strike would’ve lifted rents in every capital. It did not. And the 2026 measure keeps both concessions on new builds, which the 1985 quarantine didn’t, so the one lever that actually governs supply has been left switched on.

Where the state put its money

On 1 October 2025, the Home Guarantee Scheme was uncapped and rebranded the Australian Government 5% Deposit Scheme. The 50,000 place limit went. The income test went entirely, so eligibility no longer had a ceiling. The property price caps rose: greater Sydney from $900,000 to $1.5 million, and greater Brisbane from $700,000 to $1 million. About 70,000 buyers were expected in the first year. Treasury modelled the total effect on house prices at around half a per cent after six years.

Read that last figure again. The Commonwealth’s own analysis held that removing the volume cap, removing the income test, and lifting the Sydney ceiling by $600,000, in order to add another 20,000 borrowers on top of the 50,000 the cap already allowed, every one of them at 95 per cent, into a market at the end of a half-century run, would move prices by half a per cent over six years.

The market peaked around March 2026. The budget landed on 12 May 2026. The Opposition estimates that 61,091 first home buyers who entered between December 2025 and March 2026 now owe their lender more than the property is worth. Treat that as an Opposition estimate, because it’s one. The direction isn’t in dispute.

I have no comfort to offer those buyers that would be honest. The scheme didn’t lower anybody’s debt. It removed the deposit hurdle and left the borrower at 95 per cent, and it arrived wrapped in official reasoning, which is the whole problem. A purchase you make on your own judgement is yours, and you carry it. A purchase you make with a government programme attached arrives with the reasoning pre-installed, and reasoning is somewhere to point when it goes wrong. That converts a choice into a decision, and the people who made it stopped choosing somewhere around the announcement. It was legible at the time to anyone prepared to look at what an uncapped 95 per cent lending programme does at the top of a cycle. I told people not to touch it.

Then look at how the loss is distributed. The buyer wears the negative equity. The Commonwealth guarantees 15 per cent of each loan, so when a sale fails to cover the debt, the public covers the shortfall. The lender wears nothing. The officials who removed the income test and raised the caps wear nothing. The developers who sold into the demand the scheme created have already banked the settlement.

Terms nobody read

The negative equity is the first bill. It isn’t the last one, and the terms for the next one were drafted before anybody signed.

Help to Buy launched in December 2025, administered by Housing Australia. Under it the Commonwealth contributes up to 40 per cent of the purchase price of a new home, or up to 30 per cent of an existing one, and takes an equivalent equity share. Entry is possible on a 2 per cent deposit. Income thresholds started at $100,000 for an individual and $160,000 for a couple, rising to $110,000 and $180,000 from 1 July 2026, and they’re wage indexed annually by the same party holding the equity.

Buried in the conditions is this. If a participant’s taxable income exceeds the threshold for two consecutive years, they may be required to buy back part or all of the Commonwealth’s share, priced at market value at the time of the buy-back rather than at what the Commonwealth put in.

A household that entered on a 2 per cent deposit has no capital. So a required buy-back is funded by borrowing. Commonwealth equity converts into bank debt, the loan to value ratio jumps, and the lender ends up holding a larger interest-bearing loan against the same house. The trigger is an income threshold set and indexed by the party being bought out. Every step of it reads as reasonable on the page. You’re earning more, so you can afford to own more of your own home.

The 5% Deposit Scheme takes no equity at all, and doesn’t need any, because it holds something more useful. It holds conditions. The Commonwealth’s own guidance states that to keep the guarantee you must go on meeting obligations, including living in the property as your principal place of residence, and that if those obligations are not met the guarantee may no longer apply and the lender may require you to pay lenders mortgage insurance or other additional costs. Treasury’s own worked example puts that premium at up to $32,000 on an $800,000 purchase with a 5 per cent deposit. Capitalised onto a loan already sitting at 95 per cent.

And the guarantee only ends cleanly when the loan reaches 80 per cent of the property value. In a falling market that day recedes rather than approaches, so the conditions bind for the life of the loan. You can’t let the property out. You can’t refinance away from the participating lender without the guarantee ending. Take a job interstate, rent the place to cover the mortgage, and the protection lapses and the premium arrives.

Which leaves a 95 per cent borrower pinned to one address and one lender, while the cash rate sits at 4.35 per cent after three increases during 2026. They cannot shop. The lender knows they cannot shop.

Nobody had to collude to build that. It’s published, on a government website, and every borrower signed it. But the effect is a household whose exposure is governed by conditions written by the same party that sets migration levels, tax settings, and demand-side subsidies, which is to say the same party that determines the price the whole structure is measured against. That’s my read of the design. The record simply states the terms.

Rated for something else

None of which means rents are safe through the transition. Vacancy is sitting near 1.7 per cent and advertised rents are running about 5.9 per cent above a year ago. Rents can rise, and for some people they will. That’s a supply and demand problem, and it has a cause nobody on that panel would touch.

The accepted explanation for expensive housing is shortage, and the accepted remedy is to build. Both are half right, and the missing half is the part nobody will say.

Net overseas migration was 538,000 in 2022-23, 429,000 in 2023-24, 306,000 in 2024-25, and 301,000 across calendar 2025. The multicultural framework this country still operates was modelled in 1973 on a consistent annual net migration gain of 60,000, and its author published the projection.

A settled population growing naturally doesn’t need housing stock to expand exponentially. It needs replacement, upgrade, and household formation, which is a manageable and forecastable load. Australia doesn’t have that problem. Australia has been running the system at five to nine times its rated capacity for the past four years and calling the resulting shortfall a market failure, which is a bit like flooring a truck at nine times its load rating and blaming the axle.

I’ve set that record out in full, including what the founding document itself said about volume before the apparatus was ever built, in Hire Anyone. Import Nothing. I’m not going to relitigate it here. The part that concerns the price is arithmetic.

What the price is measured in

There’s an obvious objection to all of this and it deserves a straight answer.

A narrow band isn’t achievable in the current monetary design. A price can’t hold a level while the unit it’s denominated in is engineered to lose value. The currency is unanchored, so the issuer decides the supply. Credit creates money at the moment of borrowing rather than drawing it from any reserve. New money reaches the things people borrow against first and reaches wages last, if it reaches them at all. So a house priced in that unit has to rise in nominal terms or it’s falling in real ones.

The rising line was never a housing phenomenon. It’s a measurement phenomenon that housing displays more clearly than anything else, because housing is the most heavily borrowed-against thing most people will ever touch. Which means the honest version of the position is that a stable housing market requires a national currency anchored to something the issuer doesn’t control. That’s a longer argument and it’ll get its own treatment. Without it, a stable price is a plan to hold a level in a unit designed to move.

Positions, not businesses

Bathla isn’t a housing story.

I’ve spent close to four decades across finance, technology, hospitality, professional services, and operating roles, with private equity and venture capital as one part of that, and the same failure keeps turning up in the same shape regardless of the industry printed on the door. In a disciplined deal you underwrite the cash flow. Multiple expansion is a bonus, and you never spend it before it arrives. The moment you borrow against what a thing will be worth rather than what it earns, you’re not operating a business. You’re carrying a position, and a position decides for itself when it closes.

A developer with more than 520 subsidiaries, funded through private credit, with 15,000 dwellings in the pipeline and about 2,000 of them actually being built, was running that trade at industrial scale. So is the household that borrowed 95 per cent because the price had to keep going. So is the bank whose book only performs at the valuations it was written against. So is a country that has built its retirement expectations, its household balance sheets, and now its forward revenue estimates on the same single assumption.

So run it on your own numbers, because this isn’t a spectator argument. Which of your debts is serviced by what the business actually earns, and which is serviced by what you believe the asset will be worth?

The correction is roughly a third done. The people telling you it’s gone too far are, almost without exception, the people whose model requires it stopped. The question in front of this country isn’t how to arrest the fall. It’s whether anyone in office will let it finish, and then stand up and say that a house returning to a price one income can carry is the policy working rather than failing.

Every position described above only works if the number goes up. That’s not a reason to protect the number. It’s the reason to let it come down.