Low-Doc Arrears Hit 4.9% Against 1.01% Across the Book. That 4.9% Sits on About 0.1% of Bank Mortgages — and Your Alt-Doc File Isn’t in It

What APRA actually published — and the one figure the coverage added

Start with the regulator’s own statistics rather than the hearing, because they say something slightly different. APRA’s quarterly authorised deposit-taking institution releases for the June 2026 quarter, published on 17 September 2026, do not describe a book in distress. Non-performing residential loans — 90 days or more overdue, or otherwise impaired — sat at 1.01% of residential credit outstanding, down from 1.04% a year earlier. Residential loans 30 to 89 days past due sat at 0.54%, down from 0.66%. Residential credit outstanding grew 7.0% to $2,558.5 billion, and the share of outstanding loans at an LVR of 80% or more fell from 17.6% to 16.7%.

New lending in the quarter read much the same. ADIs funded $200.5 billion of new residential loans, up 6.8% year on year, with 29.7% written at an LVR of 80% or above (down from 30.4%) and 5.6% at a debt-to-income ratio of six times or more (up from 5.5%). Split by purpose, that DTI figure was 8.9% on investment lending against 3.7% on owner-occupied.

Then the hearing. On 29 October, APRA deputy chair Therese McCarthy Hockey told the Senate economics committee that arrears of at least 30 days were rising across borrower types.

APRA chair John Lonsdale told the same committee the authority was “closely watching the employment market which would drive a problem in that area”, and drew the distinction that matters most for a broker reading the headlines: “There is certainly mortgage stress in the community but how much of that actually translates into a non-performing loan.” Nationals leader Matt Canavan questioned him on borrowers’ capacity to keep servicing loans, and Greens senator Barbara Pocock asked for further data on notice.

The number underneath: 4.9%, and the denominator that reframes it

The figure that should genuinely change how a broker reads their own pipeline came from a separate exercise and a day earlier. Canstar’s analysis of APRA data, reported by the Sydney Morning Herald on 28 October, put arrears on low-doc loans at banks at 4.9% of outstanding low-doc loans in the June quarter, up from 4.2% in the same quarter of 2025.

Set against a whole-of-book non-performing rate near 1%, that is a different order of magnitude, and it moved in the opposite direction to the headline. The measures are not like-for-like — one is arrears on a loan type, the other is non-performing loans across all residential credit — but the direction of travel is the point.

Now the denominator, which is where most commentary stops short. The same article put low-doc loans at roughly 0.1% of bank mortgages. S&P Global Ratings analyst Erin Kitson, citing APRA, put low-doc loans at about 0.12% of total loans outstanding, down from 0.88% in 2019.

So 4.9% is a rising arrears rate on a book that has shrunk by roughly seven-eighths in six years. What is left on bank balance sheets under that label is largely a legacy slice — loans originated under older policy settings, older serviceability assumptions and an older generation of income verification. A deteriorating cohort that is also a shrinking and ageing cohort behaves differently from a deteriorating cohort that is growing.

Why that figure is not measuring the book you are building

APRA’s statistics cover authorised deposit-taking institutions. A great deal of the alt-doc business written through the broker channel today is placed with lenders that are not ADIs, and those loans do not appear in the 4.9% at all.

The last fortnight makes the point better than any argument. As reported in this masthead on 23 October, Firstmac’s alt-doc debut starts at 6.99% at 70% LVR on six months of BAS. On 25 October, Bluestone moved to apply comprehensive credit reporting across all debts while taking one year of tax returns plus a notice of assessment, lifting prime loans to $5 million under 70% LVR. Neither lender is an ADI. Every file placed with them this quarter sits outside APRA’s low-doc series.

And the single most significant policy move of the fortnight for self-employed files came from inside an ADI. Ubank, NAB’s digital lender, opened home loans to sole traders on one year of individual tax returns and a matching ATO notice of assessment. Broker Daily reported the terms on 29 October 2026, and Australian Broker reported the same change: ABN continuously active for at least two years, two years of continuous profitable trading, maximum 80% LVR, maximum 30-year term, and the most recent financial year end no more than 21 months old. Complex trust structures and construction are excluded. Ubank’s head of broker distribution, George Srbinovski, described the segment as underserved; distribution runs through more than 17,000 accredited brokers.

That matters for two reasons. The first is placement: a sole trader who would have gone to a non-bank last month may now fit a prime ADI product this month, at prime pricing. The second is measurement: those loans will sit inside APRA’s ADI data, though a single verified tax return and a matching NOA is not a low-doc loan in the 2019 sense, and nothing published so far tells us how the regulator will classify this generation of lending.

The practical conclusion is unglamorous but useful. The question for a broker is not whether 4.9% is their number — it almost certainly is not. It is what their file notes will say when somebody builds the cohort that does include their files, two or three years from now, and goes looking for self-employed stress.

What it does tell you: a borrower-type signal, and a lagging one

Kitson’s framing is the one worth carrying. Low-doc loans were more vulnerable to arrears without presenting a systemic risk, she said, because the cash flows of those borrowers are “more exposed to deteriorating economic conditions, including rising interest rates”. She added that rising cost-of-living pressures were “not yet manifesting in material arrears increases”, and that mortgage arrears are “one of the last places for financial pressure to manifest”.

That last observation is the operationally important one, and it is easy to skim past. Arrears is a lagging indicator. The self-employed client whose trading revenue softened in the September quarter does not miss a mortgage payment in the September quarter. They draw down the offset, then the redraw, then the business card, then the ATO payment plan. By the time the loan is 30 days down, the sequence has been running for a year. Which means a broker watching arrears data to decide when to call their self-employed back book is watching the wrong instrument.

Canstar’s data insights director Sally Tindall put the rate transmission plainly in the same article:

The rate backdrop is unusually dense. The cash rate sits at 4.60% after four increases in 2026: to 3.85% effective 4 February, 4.10% on 18 March, 4.35% on 6 May and 4.60% on 30 September. The board’s next decision lands on 3 November, and market pricing has converged on a fifth rise that would take the rate to 4.85% — its highest in 18 years. The Adviser reported on 28 October that all four major banks were tipping a November increase. Unemployment reached 4.7% in September, a five-year high.

Nothing in this article turns on that decision going one way or the other. But the arithmetic of a higher base is worth stating: Canstar’s research puts non-bank low-doc advertised pricing from about 7%, and only with 45% equity, rising to about 12.29% where the deposit is smaller. A client carrying a rate that starts with a seven feels each 25-point move on a materially higher base than a client on a prime variable.

Where RG 273 actually bites on an alt-doc file

The best interests duty exposure on an alt-doc file is usually not about documentation. It is about price.

ASIC’s Regulatory Guide 273 says at RG 273.51 that it generally expects the cost of a credit product “to be a factor that mortgage brokers should prioritise”. At RG 273.54 it is more direct: “Any situation where a higher cost loan is recommended will need to be supported by evidence.” The guide also leaves brokers real room — at RG 273.56 it acknowledges that “the product with the lowest interest rate is not necessarily the lowest cost option for all consumers.”

Read together, that is not a rule against alt-doc. It is a rule about the record. An alt-doc recommendation is unremarkable where the file shows why a cheaper full-doc route was closed. An alt-doc recommendation where the file says only that the client is self-employed is a thinner note this month than it was last month — precisely because Ubank now accepts one individual tax return plus an NOA at 80% LVR, and Bluestone is doing something comparable in near-prime. The set of files where “no full-doc option existed” is true has narrowed, and it is narrowing in public, in the trade press your reviewer also reads.

On the forward-looking element, RG 273.48(d) directs attention to “reasonably foreseeable changes to the consumer’s personal circumstances and financial situation”. Four cash rate increases inside a calendar year, with a fifth decision scheduled and widely anticipated, is not an unforeseeable event.

This is general information about what the guide says, not advice about how it applies to any particular file or licence. How these obligations operate in practice depends on your credit licence, your aggregator’s process and the facts of the file. Confirm the application with your licensee or compliance adviser before changing a process.

What to review this week

What to watch next

The 3 November decision and the pricing gap. A fifth increase widens the dollar gap between prime and alt-doc on the same loan size, which is the gap RG 273.54 asks you to evidence.

APRA’s answers on notice. Lonsdale told the committee that more first home buyers are moving into the 90% and 95% LVR categories since the 5% deposit scheme expanded in October 2025, while noting the data was incomplete. Senator Pocock asked for figures on notice, so a fuller picture should reach the committee in writing.

The 30–89 day bucket in the next quarterly release. That is where McCarthy Hockey’s “ticking up” would first appear in published statistics. At 0.54% in June it was still falling year on year.

Whether more ADIs follow Ubank. Each one that accepts a single tax return at prime pricing narrows the set of files where an alt-doc rate is the only option — and therefore widens the set where the file note has to do the work.

Sources: APRA, Quarterly authorised deposit-taking institution performance and property exposure statistics, June 2026 quarter, released 17 September 2026; evidence to the Senate economics committee, 29 October 2026, as reported by The Nightly (29 October 2026) and The Adviser (2 November 2026); Canstar analysis of APRA data and interviews reported by the Sydney Morning Herald, 28 October 2026; Broker Daily and Australian Broker on Ubank’s sole trader policy, 29 October 2026; The Broker Times, 23 and 25 October 2026, on Firstmac and Bluestone; ASIC Regulatory Guide 273, Mortgage brokers: Best interests duty; RBA cash rate statistics; ABS Labour Force, Australia, September 2026.