Arrears Fell to 1.01% and LVRs Above 80% Eased to 29.7%. The Same APRA Release Shows a Record-Thin Deposit Tail

What APRA actually published

On 17 September 2026, APRA released its Quarterly ADI Performance and Quarterly ADI Property Exposures statistics for the quarter ending 30 June 2026. On the headline measures, the picture is one of a system that spent the year getting modestly safer.

New loans funded totalled $200.5 billion, up 6.8% year-on-year. Of that new lending, 29.7% was written at a loan-to-valuation ratio of 80% or above — down 0.72 percentage points from 30.4% a year earlier. The share of existing loans sitting above 80% LVR fell further, down 0.9 percentage points to 16.7%, which is what you would expect after several years of price growth lifting older loans down the LVR curve.

Credit quality improved on both of APRA’s measures. Loans 30 to 89 days past due fell to 0.54% of residential exposures, from 0.66% a year earlier. Non-performing loans fell to 1.01%, from 1.04%. Debt-to-income concentration barely moved in aggregate: loans written at six times income or more made up 5.6% of new lending, against 5.5% a year ago.

If you stopped reading there, the reasonable conclusion would be that borrower stress peaked and is now easing, and that lenders quietly tightened the top of the LVR range. That conclusion is not wrong. It is just incomplete in a way that matters commercially.

The same release, read as a distribution

Averages hide shape. When you look at where the growth in that $200.5 billion actually came from, the book is not tightening evenly — it is thinning at both ends while the middle stays calm.

At one end, the deposit floor. Australian Broker, drawing on Canstar’s analysis of the same APRA data, reported on 18 September that new owner-occupier loans written with a deposit of 5% or less reached 4.31% of new lending — described as a record — and that roughly $15.6 billion of such lending has been written since October 2025. Those figures appear in that reporting rather than in APRA’s own summary commentary, so treat them as Canstar’s read of the underlying tables rather than an APRA headline.

Canstar data insights director Sally Tindall put it this way in that report: “The number of borrowers getting into the property market with barely any skin in the game has surged yet again, at the same time the housing market is shifting into reverse.”

At the other end, investors. The investor share of new lending rose to 35.6%, up from 34.1%, while the owner-occupier share fell to 61.9% from 63.6%. And investors carry almost all of the DTI concentration: 8.9% of investor loans were written above six times income, up from 8.7%, against just 3.7% for owner-occupiers, which was essentially flat. The aggregate DTI number looks calm because owner-occupiers are diluting an investor cohort that is not.

Two more lines round out the shape. Interest-only lending sits just under a quarter of new loans — Australian Broker put it at 24%, up from 21% a year earlier. And roughly 5.8% of new loans were funded as exceptions to serviceability policy: about one loan in every 17.

Why “arrears fell” and “arrears rose” are both true

This release produced trade headlines pointing in opposite directions on arrears, and it is worth understanding why, because brokers will be asked about it.

APRA’s ratio fell: non-performing loans went from 1.04% to 1.01% of credit outstanding. But the denominator grew. Total residential credit outstanding reached $2,558.5 billion, up 7.0% year-on-year. Run the arithmetic on APRA’s own two figures and the dollar balance moves the other way: 1.01% of $2,558.5 billion is roughly $25.8 billion, while 1.04% of a book around 7% smaller a year earlier is roughly $24.9 billion. That is close to a billion dollars more in non-performing housing debt, sitting inside an improving percentage.

Both framings are defensible. They answer different questions. The ratio answers “is the system’s credit quality deteriorating?” — and the honest answer is no, not on this data. The dollar balance answers “are there more households in serious trouble than last year?” — and on the same data, the answer is yes.

Offsets move before arrears do

The most forward-looking line in the release is not an arrears number at all. It is the offset data.

On APRA’s figures, offset balances as a portion of total credit outstanding slipped to around 13.3%, off a peak near 13.9% in late 2025. Australian Broker reported the dollar detail: total offset balances of $340.5 billion, still up 12.8% year-on-year, but down $8.6 billion over the June quarter.

That combination — up strongly on the year, down sharply in the quarter — is the signature of households drawing on a buffer they spent two years building. Offset balances are discretionary. They fall when a household decides, month by month, that it needs the cash more than it needs the interest saving. No lender reports that decision, no credit file records it, and it happens long before a payment is missed.

For brokers this is the rare leading indicator you can actually observe. You cannot see a client’s arrears status. On many files you can see, or can reasonably ask about, their offset balance at review time. A client whose offset has gone backwards for two consecutive quarters is telling you something that will not show up in APRA’s data for another year.

Why this lands in the broker channel first

With the broker channel originating roughly four in five new residential loans on the MFAA’s most recent market-share reporting, any cohort defined by “recently written” is, by arithmetic, mostly broker-written. That is not a criticism of the channel — it is a description of who holds the relationship when a recent, thin-deposit, high-LVR loan runs into a softer market.

The practical problem is mobility. A borrower who settled at 95% or higher in the past twelve months, in a market where Australian Broker reported Cotality’s estimate that around one in two homes bought in Sydney and Melbourne over the past year would now sell below their purchase price, is unlikely to have the equity to refinance, restructure, consolidate or top up. Sydney was reported down 7.7% and Melbourne down 6.9% from their April peak.

That client is not necessarily in trouble. They may be perfectly comfortable. But they are immobile: they cannot be repriced by moving them, they cannot be helped by a cash-out, and if their circumstances change, your options for them are limited to what their existing lender will agree to. A trail book with a growing immobile cohort behaves very differently from one where most clients can be refinanced on request — both for your revenue forecasting and for the kind of conversations you will be having in eighteen months.

What to review this week

This is a segmentation exercise, and it uses data you already hold in your CRM. It should take an afternoon, not a project.

What you do with the flagged list matters more than the list. The useful move is a proactive annual review contact — not a refinance pitch, which for much of this cohort will not be available anyway. It is a check-in that establishes the client’s current position, surfaces the offset trend, and puts you in the conversation before a problem is urgent. Brokers who make that contact in a soft market tend to be the ones still holding the relationship when the market turns and the refinance does become available.

The conversation this changes

Australian mortgage brokers operate under a best interests duty, introduced into the National Consumer Credit Protection Act 2009 and the subject of ASIC’s Regulatory Guide 273. Responsible lending conduct sits alongside it under ASIC’s Regulatory Guide 209. This article is general information and not a statement of what either instrument requires on any particular file — your licensee or aggregator compliance team is the right source for that, and the obligations that attach to your credit licence will be specific to your arrangements.

What the data does raise is a practical question about the quality of the record you leave behind. When a loan is written at a very high LVR, on interest-only terms, or as an exception to a lender’s serviceability policy, the reasoning behind that recommendation is doing real work. A file that records why a particular structure suited that client’s circumstances at that time — and what alternatives were considered — is simply a more robust file than one that records only the outcome. That is true regardless of what any regulator does next, and it is most valuable precisely in the cohort this release identifies as growing.

The related conversation is with the client. A borrower entering at a 5% deposit in a falling market should understand, at the point of advice, that their ability to move lenders will depend on the valuation holding — not just on their income. That is not a reason to decline the business. It is a reason to have the conversation while the client is still choosing, rather than in eighteen months when they ring asking why nobody will refinance them.

What to watch next

Three things will tell you whether the tail identified in this release is a temporary feature or a trend. First, the September 2026 quarter data, due from APRA in December, and specifically whether the thin-deposit share holds above 4% or retreats. Second, whether the 30–89 day past due measure — the earlier of APRA’s two arrears series — turns up before the non-performing series does, which is the usual order. Third, whether offset balances fall again in the September quarter, which would turn a single quarterly drawdown into a pattern.

None of those is a reason to change how you write business tomorrow. All three are reasons to know, now, which of your files would be affected if they move the wrong way — because the work of finding out takes an afternoon today and considerably longer under pressure.

The bottom line

The June 2026 quarter is not a bad set of numbers. The Australian mortgage book is performing better on APRA’s credit quality measures than it was a year ago, and high-LVR lending as a share of the whole is going down, not up. But a system average is a poor proxy for a broker’s portfolio, and this particular average is being held down by a large, seasoned, low-LVR middle that most brokers do not hold much of.

What brokers hold is recent business. Recent business is where the thin deposits, the interest-only structures and the serviceability exceptions are concentrated, and it is the only cohort for which the market has moved further than the loan has paid down. Read the release for what it says about the system, then go and read your own settlement data for what it says about you. They will not tell you the same story, and the second one is the one you can act on.

Sources: APRA, Quarterly Authorised Deposit-taking Institution Performance Statistics and Quarterly Authorised Deposit-taking Institution Property Exposures Statistics, June 2026 quarter, released 17 September 2026 (apra.gov.au). Deposit-band, interest-only, serviceability-exception and offset dollar figures, the Sally Tindall quotation, and the Cotality price estimates as reported by Mina Martin, Australian Broker, “Low deposit lending hits record high as mortgage arrears climb”, 18 September 2026. Broker channel market share per MFAA market-share reporting. Dollar estimates of non-performing balances in this article are The Broker Times’ own arithmetic applied to APRA’s published share and credit-outstanding figures, and are approximate.