The Price Floor Has a Postcode Problem

Replacement cost only holds prices up where you can actually build. In the middle ring you can't — and at the fringe, where you can, the bank has already said no.

I wrote last week about Ray White's modelling that Brisbane needs a 32% price fall to bring investors back under the quarantined negative gearing rules. That figure is routinely quoted without its qualifier: it assumes rents do not move at all. On that basis Brisbane needs around 32% and Sydney around 28%, measured against a 5.15% minimum yield hurdle.

The same research does not expect prices to move anything like that far, and the reasoning is worth setting out in full, because it is more careful than the summaries of it.

What is expected instead is a combination of stronger rents and softer prices, lifting yields gradually rather than through a single sharp price-only correction.

On how far prices fall, the argument rests on two things: replacement costs rising, and housing delivery already running well short of target. On that basis a downturn comparable in depth and duration to the GFC is considered unlikely.

And it does not ignore credit. The framing of this cycle opens with higher interest rates having reduced borrowing capacity. The constraint is named, and the conclusion is still that the floor holds.

That is the argument standing between the 32% and the market, and it should be taken at full strength.

The cost base is not going back

The ABS output price index for house construction is up more than 50% nationally since December 2019. That is not a cyclical spike that unwinds when timber prices settle. Labour, compliance, insurance and finance costs have all repriced, and none of them are going back.

On the state breakdown, Queensland tracks above the national line — roughly 1.6 against 1.5.

Which means the capital needing the largest adjustment also carries the highest floor of anywhere in the country. That is not a contradiction — it is precisely why the argument expects most of the movement to come through rents rather than prices.

The assumption underneath it

So the disagreement is not about whether borrowing capacity has fallen. It has, and the argument says so. It is about what happens next.

A floor only holds if buyers can bid at it.

Replacement cost sets what a developer needs to recover to justify building. It is a supply-side floor. It says nothing about what a bank will lend to the person standing in front of the property.

When those two numbers agree, the mechanism works exactly as described. When they don't, the market clears at the lower one — because a developer's spreadsheet has never once cleared a transaction. The bank's serviceability calculation does.

Construction costs have risen permanently and that much is correct. The open question is where the demand forming that floor actually stands, and whether it can still pay when it gets there.

The floor only exists where both products exist

The mechanism needs established stock and new stock to be substitutes. A buyer looks at the price of building new, decides established is better value, and that switch is what holds established prices up.

That only works where both options are actually on the table.

In the middle ring — Ashgrove, Bulimba, Camp Hill — there is no new supply to substitute away from, because there is no land to put it on. Replacement cost can rise another 50% and it will not change what someone pays in those suburbs, because nobody is choosing between an established home there and a new build there. The choice does not exist.

Replacement cost has nothing to say about a suburb where you cannot build.

Where the two products do sit side by side is the outer growth corridors. In places like Warner in North Brisbane, a new house-and-land package genuinely competes with the established four-bedroom two streets away, and it carries a premium over it. That premium is the substitution mechanism, live and working exactly as described.

So the floor is real, and it is local. It operates at the fringe.

Which is the worst possible place for it

The fringe is where the marginal buyer stands. First home buyers, upgraders stretching to get in, investors chasing yield at the affordable end. It is the most credit-sensitive part of the market by a distance, and it is the first place a serviceability calculation starts saying no.

So the floor is strongest precisely where the buyer is weakest. The mechanism that is supposed to hold prices up depends on people bidding at a level the bank has already declined to fund.

That is the structural problem. And there is a way to see whether it has started.

The exempt segment is the one that slumped

Here is the test the policy set up for itself.

New builds were not touched by the negative gearing and CGT changes. Established property bought after 12 May 2026 gets the quarantine; a new build is fully exempt, whenever it is bought. The reform deliberately pushes investors toward new supply.

So if tax were the thing driving investors out of established housing, new home sales should be absorbing them.

Both of the country's largest residential developers reported full-year accounts on 20 August, and both said the opposite.

Stockland settled 8,902 lots in FY26 — a company record, well above its own 7,500 to 8,000 guidance. It has guided FY27 down, to 7,300 to 8,300, and told the market inquiries for new home sales slumped over the past three months. On when sentiment turns: only once rates are cut, prices fall, or wages start rising more strongly. On timing, its read was that the cycle usually runs 13 to 18 months and "we are only six months in".

Mirvac settled 2,130 lots, within guidance but under analyst expectations, and described sales as moderating with inquiries slowing in the three months to 30 June. Campbell Hanan put investor purchases below their long-term average of 30%, and said the market needs to stabilise — specifically, that buyers want to see auction conversion rates recover — before investors return.

The one segment the tax change protects is the one where inquiries fell.

That is the first hard evidence from the supply side, and it comes out of company accounts rather than commentary. It also isolates the cause. A tax explanation cannot account for a slump in the segment the tax exempts. A credit explanation can.

Mirvac is also forecasting construction cost escalation of 4.5% in New South Wales and around 4% elsewhere this financial year. The cost base is still climbing while inquiries fall — which is the squeeze in a single line.

Neither company is forecasting collapse, and it would be dishonest to imply otherwise. Mirvac has guided settlements up. Stockland's range still overlaps its prior guidance. What they are describing is a softer year off strong pre-sales, in a cycle with a year or more left to run.

And the trade-down segment has stopped moving

Units are the trade-down segment — where buyers and renters go when houses get away from them. If affordability is binding anywhere, it shows there first.

In the June quarter, Brisbane house rents rose 2.9% to a record $700 a week. Unit rents did not move at all, flat at $660, ending six consecutive quarters of growth. Both segments sat on the same 0.6% vacancy, equal-lowest June on record — so it is not a supply story.

One quarter is not a trend. But it is the first quarter the two segments have parted under identical conditions, and the cheaper one is the one that stopped.

What it means for the floor

The replacement cost argument needs displaced demand to land somewhere below and hold prices up from underneath.

At the fringe, it lands on buyers the bank has already declined. In the middle ring it does not land at all, because there is nothing to land on.

Construction costs still tell you what a dwelling ought to cost to reproduce. In a suburb where you cannot build, that is a number with nothing attached to it.

And the qualifier cuts both ways. The 32% is the no-rent-growth corner of the model — the number you get if rents do nothing at all. Let rents rise and it falls away quickly: nationally, a 10% rent rise takes the required price adjustment to around 16%, and a 20% rise to around 8%.

Which is the real point. The rent side is carrying almost all of the load in every scenario where prices hold up.

So a segment where rents have stopped moving is worth more attention than any headline percentage.

What this changes for a principal

Two things, and neither is abstract.

If you are pricing a vendor off replacement cost, you are pricing off the wrong number. "They couldn't build it for that" is a comforting line in an appraisal and it will not survive contact with a buyer's finance approval. The number that clears a sale is what a bank will advance, and that is set by serviceability, not by the cost of bricks.

If you hold a rent roll weighted to units, last quarter your income growth stopped. Management fee income is a percentage of rent collected. A flat quarter on units is a flat quarter on that portion of your fee base — and if you borrowed to acquire that rent roll, the debt behind it is repricing while the income behind it isn't.

Neither of those shows up on a P&L. Both show up in a forward schedule, if there is one.

No principal works through their rent roll position in a comment section. If you are quietly running that number, send me a DM.

John King is a Fellow of the Institute of Public Accountants and an Authorised Credit Representative. Peer Group Advisory works with boutique real estate agency principals on forward cash flow, structuring and acquisition funding.

Peer Group Advisory