The Losses Have an Address

Spring's third weekend printed 54.0% on 1,832 auctions (1,354 reported: 732 cleared, 622 uncleared). The book grew 14.9% on the week to its biggest since 21 June, and the clearance rate gave back 4.5 of the 5.8 points it gained the week before. Volume and price moved in opposite directions, which is the ordinary arithmetic of a spring selling season inside a correction. More stock arrives, buyer numbers do not rise with it, and the marginal property fails to clear.

The composition weakened with the headline. 445 homes sold under the hammer against 275 prior to auction and 12 after, so the hammer's share of cleared sales held at 61%, but 371 passed in (27.4% of reported, from 25.3%) and 251 were withdrawn before the day, lifting the withdrawal rate to 13.7% of everything scheduled from 11.9%. Withdrawals are the quiet half of a clearance rate. A vendor who pulls the property never appears in the denominator, so the published figure flatters a market where withdrawals are rising.

Melbourne: the pre-Grand Final push

Melbourne scheduled 900 auctions, its largest book since 21 June, and cleared 56.3% (714 reported, 402 cleared). The volume has a date behind it. The AFL Grand Final is Saturday 26 September, and agents routinely clear their books the weekend before rather than compete with it. Read next weekend's Melbourne print as a seasonal artefact and nothing else. Within this weekend's numbers, 281 sold under the hammer and 213 passed in, with 116 sold prior and 99 withdrawn (11.0% of scheduled). Clearance fell 7 points from 63.3%, which is the largest single-week drop of any capital, and the fall came on 187 extra auctions. At 713 auctions Melbourne cleared 63%; at 900 it cleared 56%.

Sydney: the negotiating table moved off the auction floor

Sydney scheduled 577 and cleared 54.2% (417 reported, 226 cleared), down 5.4 points. 130 homes sold before their auction against only 92 under the hammer, and a further 120 were withdrawn, 20.8% of everything scheduled and up from 16.5% a week earlier. More than a fifth of Sydney vendors decided, in the days before their scheduled auction, that a public test of price was not in their interest. Combined with a median 45 days on market (32 a year ago) and a median vendor discount of 4.5% (3.2%), the picture is a market where price is settled in private negotiation and the auction increasingly serves as marketing.

Brisbane, Adelaide, Canberra, Perth

Brisbane slipped to 37.5% (120 reported, 45 cleared) with 57 passed in. The pass-in share has not been below half since June, and total listings sit 53.3% above a year ago, the largest listings build-up of any capital. Adelaide posted 58.2% (55 reported, 32 cleared), the best of any capital on a meaningful sample, with 27 sold under the hammer and only 7 withdrawn from 113 scheduled, the lowest withdrawal rate in the country and consistent with a market 1.6% off its peak rather than 7.1%. Canberra rose to 57.5% from 46.9%, the largest weekly gain of any capital, on 40 reported. Perth returned 8 results at 50.0%; it is a private-sale market and the auction read carries no information.

The year-ago comparison, and the two series

Two organisations count this weekend and they do not count the same thing. The figures above are Cotality's preliminary series, which is what this newsletter has tracked since Edition 1. My Housing Market's separate read was 50.6% on 2,086 auctions, down from 51.9%, against 72.3% on the equivalent weekend a year ago; its per-city reads were Sydney 54.0% (796 auctions), Melbourne 59.3% (913), Brisbane 30.8% (177), Adelaide 53.9% (123) and Canberra 55.2% (77). The two series differ in which auctions they capture and when they close the count, so the levels are not comparable, though both series fell this week.

On volume the year-ago comparison is unambiguous, and it is the most important number in this report. Cotality scheduled 2,647 auctions on the equivalent weekend in 2025 against 1,832 this weekend, a fall of 30.8%. Set that against the other side of the market. Advertised stock across the combined capitals is 24.4% higher than a year ago, and the national total of 139,167 is up 18.1% and above its five-year average for the first time in 2026. The capitals figure is the like-for-like one, because the auction count is a capital-city measure. (Listings are Cotality's four weeks to 6 September; the auction count is this weekend, so the two are not the same window.)

Those two facts together change how the clearance rate should be read. There is more property for sale in Australia than there was a year ago, and far less of it is being taken to auction. The auction channel is not a fixed window onto the market; it is a shrinking one, and what remains inside it is self-selected. A vendor who expects a contested result books an auction, and a vendor who does not lists by private treaty or, as 251 did this weekend, withdraws before the day. So 54.0% is struck on a smaller and more favourably composed sample than last year's number was, which means the year-on-year fall in clearance understates the deterioration rather than overstating it.

The absolute count makes the same point without any percentages: 732 homes cleared at auction across the capitals this weekend. A year ago the combined capitals scheduled almost 45% more auctions into a market clearing in the low 70s on My Housing Market's series. Finals typically settle 3 to 5 points below the preliminary print, so expect this weekend to finalise near 50%.

Tasmania did not report auction results this weekend. The combined-capitals cleared count (732) equals the sum of the six reporting cities exactly, confirming no Tasmanian contribution to this week's total.

The cash rate has been 4.35% since the Board held for a second consecutive meeting in August, after three hikes between February and May 2026. The 28-29 September meeting is the live one, and by the close of this week markets were pricing roughly an 80% probability of a 25-basis-point move to 4.60% on the Tuesday.

The professional forecasters split four to three, and they disagree only on the month. NAB, UBS, Morgan Stanley and Deutsche Bank expect the move at the September meeting. ANZ, CBA and Westpac expect it in November, their argument being that the Board will want the September-quarter CPI and a full set of revised forecasts before acting. No major forecasts a hold through both meetings. For a borrower the difference between September and November is six weeks of settlement timing, and every major expects the next move to be up.

The inflation backdrop is why. Headline inflation eased to 3.5% in July from 3.8% in June, but the trimmed mean held at 3.6%, above the top of the 2-3% target band. A headline figure falling on fuel and electricity while the underlying measure holds is the configuration that historically keeps a central bank hiking, because the second number is the one the Board treats as signal. The August monthly CPI lands around 30 September, after the meeting.

What 4.60% does to the arithmetic

This year's 75 basis points have already added about $350 a month to repayments on the average new owner-occupier mortgage of roughly $735,000, and cut the borrowing capacity of a median-income borrower by 7.0%, more than $53,000. Another 25 points lifts a lender's serviceability assessment from roughly 9.4% toward 9.65% and trims capacity a further 2 to 3%.

Borrowing capacity is the transmission mechanism between the cash rate and the price you can pay. It moves prices without any distressed selling. If the marginal buyer at auction can borrow 3% less next month, the marginal bid falls by something close to 3%. That is visible in this week's clearance data, in the 27.4% pass-in share, and in a median capital-city vendor discount that has widened to 4.2% from 3.0% a year ago.

The credit side of the market is already responding without waiting for the meeting. Investor loan commitments fell 10.2% by value in the June quarter, the largest quarterly fall since September 2022, and the investor share of new lending dropped to 38.0% from above 40% earlier in the year. Owner-occupier commitments fell 1.9%. Mortgage stress reached an 18-year high of 32.5% of borrowers in July on Roy Morgan's measure.

How to position

Assess any purchase you are contemplating at 4.85% rather than today's 4.35%. If the deal only works at the current rate, it is a deal with no margin for the outcome that four of seven major forecasters expect within two weeks. The corollary matters too: the rate cycle that follows a hike into a weakening economy is usually shorter than the one that follows a hike into a strong one, and the banks that forecast a November rise also pencil cuts from around August 2027. That is the earliest date on which credit stops subtracting from borrowing capacity and starts adding to it, and it is the date around which a three-to-five-year hold should be framed.

SQM's August release landed mid-month and national vacancy held at 1.3% for a third consecutive month, on 41,039 vacant dwellings. The steady headline conceals two things moving underneath it.

The first is that the stock of vacant rentals is 8.7% higher than a year ago, when the rate was 1.2% on 37,742 vacancies. A flat rate with a rising count means the rental stock is growing at roughly the pace vacancies are, so tenants have more choice in absolute terms than the percentage implies.

The second is that the loosening is concentrated. Canberra rose to 2.1% from 1.8%, Melbourne to 1.8% from 1.7%, and Darwin to 0.4% from 0.3%, while Sydney (1.7%), Brisbane (0.9%), Perth (0.6%), Adelaide (0.6%) and Hobart (0.6%) were unchanged. Five of eight capitals remain below 1%, a level at which a tenant in Brisbane, Perth, Adelaide, Hobart or Darwin has essentially no negotiating position. Canberra and Melbourne are a different market operating under the same national headline.

Rents stopped rising

Nationally, advertised rents were flat over the month for the first time this cycle, with the annual rate at 7.3%. The city detail: Sydney -0.6% over 30 days (houses -1.2%, to about $1,129 a week), Canberra -1.6% (houses -3.2%), Melbourne -0.1% (units -0.6%), against Brisbane +0.5%, Adelaide +0.4% and Perth +0.2%.

Asking rents on new listings lead the market. Cotality's whole-of-stock series still recorded 5.7% annual growth to August, and the CPI rents component, which measures rents actually paid across all existing leases, was 3.6% in July. Those three numbers are not in conflict; they are the same phenomenon at three different lags. The leading series has stalled, the stock series is still rising, and the lagging series will keep reflecting earlier increases for a year or more as leases roll. That ordering is what you would expect near a turning point in rents.

One qualifier belongs on it. Winter is seasonally the softest period for advertised rents, so a flat August is suggestive rather than conclusive. The test is dated. Asking rents normally firm from October into the December-January leasing season. If that lift does not appear in Sydney, Melbourne and Canberra, the rental upswing in those cities is over.

What it means for the yield case

The national gross yield is 3.79%, the highest since September 2019, up from a 3.5% low in early 2026. Almost all of that repair has come from the denominator, with values 3.6% below their March peak. In Sydney, on a 3.3% gross yield with asking rents falling month on month, a yield-expansion argument is a restatement of the price fall. In Brisbane (0.9% vacancy, rents +7.7%), Perth (0.6%, +7.1%) and Hobart (0.6%, +10.4%), rent growth is contributing to the yield, and the income case for a purchase there survives a flat price forecast. Adelaide is the exception among the tight capitals, with 0.6% vacancy but only 3.4% annual rent growth, so its 3.6% yield has to be earned on entry price.

The losses have an address

Resale profitability turned this quarter, for the first time in this cycle.

Across roughly 94,000 residential resales in the June quarter, 95.4% sold for more than the owner paid, down from 96.1% in March, which was a 21-year high. The median nominal gain fell from $378,000 to $371,000 and the median loss rose $1,000 to $45,000. Read only those figures and you would conclude the downturn has barely touched the selling side. That conclusion would be wrong, because the national average is an average of two very different populations.

The two variables that decide it

The first is how long the property was held. Profitable resales had been owned a median 9.1 years. Loss-making resales had been owned 8.1 years, and loss-making houses just 4.4 years against 9.3 years for profitable ones. Length of hold matters more here than choice of market. Nine years of compounding growth absorbs a 3.6% peak-to-current fall; four years does not.

The second is what was bought. Houses cleared a profit on 97.8% of resales at a median gain of $435,500. Units cleared 90.5% at $251,000. The gap between the two median gains has widened from 21.2% five years ago to 36.5% now, which reads as a structural divergence rather than a cyclical one.

Five council areas, 39% of the losses

Combine the two variables and the 4.6% resolves into something specific. Sydney and Melbourne units accounted for 83.3% of the value of all unit resale losses nationally. More than one in five Melbourne unit resales, 20.8%, sold below the previous purchase price. In Sydney, 11.4% did. Five local government areas (Melbourne, Parramatta, Stonnington, Port Phillip and Sydney) generated 39% of national unit losses between them.

These are the postcodes that absorbed the 2016-2019 apartment construction wave and the 2021-2022 off-the-plan settlements. Supply landed, rents were soft through the pandemic, and owners who sold inside a decade sold into a stock overhang. The contrast with the mid-sized capitals is stark: Brisbane returned 99.8% profitability at a median gain of $525,000, Adelaide 98.9% at $480,400 and Perth 98.8% at $470,000. Melbourne, across all dwelling types, returned 89.0% at $278,000. Regional Australia outperformed the capitals on the profitability measure, 97.5% against 94.1%, though the capitals produced the larger median gain at $415,000.

Why the next print will be worse

The national Home Value Index fell 1.5% over the three months to June, the window this data covers. It has fallen a further 1.9% since, and values now sit 3.6% below the March peak. The September-quarter report will capture sales made into that second leg, and the same two variables will decide who lands on the wrong side of it: short holds, and units in the two cities where supply has been heaviest.

The seller-side evidence already points that way. The median vendor discount across the capitals widened to 4.2%, from 3.0% a year ago and the largest in at least two years, and median days on market stretched to 37 from 25. Sydney is at 45 days and a 4.5% discount, Perth at 22 days and 4.4%, Brisbane at 35 days from 19. Vendors are accepting less, and waiting longer for it.

The investor takeaway

This is an argument against one specific combination rather than against units, or against Sydney and Melbourne. That combination is a short intended hold, in a dwelling type with elastic supply, in a market where that supply has already landed. Buy that combination and the odds of a nominal loss run between one in nine and one in five. Buy a house you intend to hold nine years and, on a 21-year record, they run near one in fifty.

Three practical tests follow. First, can you hold through a full cycle, which on this evidence means closer to a decade than to five years? If not, dwelling type matters far more than suburb selection. Second, in any unit purchase, screen the building era and the strata history before you look at the suburb's yield table, because the losses cluster by vintage and body corporate rather than by postcode alone. Third, if you are selling, remember that the 95.4% is the record of a market that was rising until March. Price against the current quarter instead.

The week ahead

Three dated items. Saturday 26 September is the AFL Grand Final, so Melbourne's auction volume will collapse and the national print will be thin and uninformative; ignore it. Tuesday 29 September the RBA Board decides, with markets pricing roughly 80% for a move to 4.60%. Around 30 September the ABS publishes the August monthly CPI, which lands after the decision rather than before it. Cotality's September Home Value Index follows on 1 October and will show whether the sixth consecutive monthly fall came in above or below August's 0.9%.