Bank Branches Fell 3.0% and the Regional Decline Was the Smallest on Record. The Moratorium Holding It There Runs Out Mid-2027 for NAB and ANZ

What APRA actually published

APRA’s Points of Presence collection is a once-a-year count of the physical banking channels Australian ADIs operate — branches, ATMs, EFTPOS facilities and other service points — categorised by remoteness under the Australian Statistical Geography Standard. The 14 October 2026 release covers the year to 30 June 2026, and the accompanying data file now runs from June 2017 to June 2026. The next edition is scheduled for October 2027.

Two sentences in the release do the work. The first: “The number of bank branches nationwide fell 3.0 per cent between June 2025 and June 2026.” The second: “Regional and remote areas experienced the smallest fall recorded since this data series began (1.7 per cent).” APRA adds that the decline is slower than in previous years, and that a government agreement with the major banks on a moratorium on regional branch closures partly explains the regional result.

That last clause is the one worth sitting with. APRA is not presenting the regional number as a market outcome. It is presenting it, in part, as a policy outcome — and policy arrangements have end dates in a way that market trends do not.

The counts behind the percentage

APRA’s release gives percentages rather than totals. The counts in circulation this week come from Canstar’s analysis of the same APRA file, as reported by Australian Broker on 15 October. Because these figures have appeared through one analyst’s read of the data rather than in APRA’s release text, they are worth attributing as such when you quote them to a client.

On Canstar’s numbers, the network lost a net 97 branches over the year to 30 June 2026, taking the national total to 3,108. That is a five-year net fall of 1,215 branches, or 28% since June 2021. Measured against the 5,694 branches recorded at the start of APRA’s series in June 2017, the network is roughly 45% smaller.

Bank-owned ATMs fell by 184 over the year to 4,959, which Canstar describes as a network more than a third smaller than five years ago. CBA made the deepest ATM cut at 45 machines; NAB was the only major to add ATMs, with 21, while still operating the fewest of the four.

The closures moved to the cities

The most useful thing in the data for a broker is not the headline fall. It is the geography of it, which has inverted from the story everybody learned to tell between 2020 and 2023.

Canstar’s split puts net 72 of the year’s 97 closures in major cities, and net 25 outside them. The non-metropolitan figure has collapsed from 122 in FY2022 to 29 in FY2025 to 25 in FY2026. Among the majors, Westpac recorded the largest reduction at 17 branches, down to 475, with only one of those in a regional area; CBA was down 14, NAB down 4, and ANZ was unchanged at 284, having shifted one branch from a major city to an outer regional area. Nearly all the major-bank reductions, on Canstar’s read, were metropolitan.

The moratorium has a date on it

APRA names the moratorium as part of the explanation for the regional result but does not set out its terms. Canstar’s reporting of the arrangements gives the timeframes: CBA’s commitment extended to December 2030, Westpac’s running through 2030, and NAB’s and ANZ’s expiring in mid-2027.

Mid-2027 is not a distant horizon. It is inside the life of almost every pre-approval you will write between now and Christmas, and well inside the first fixed term on a loan settling this quarter. Tindall’s framing, as quoted by Australian Broker, is the right one to carry into a client conversation: “However, this problem has been paused, not solved.”

None of that is a prediction that NAB or ANZ will close regional branches in July 2027. It is simply the difference between a commitment with a stated end date and a commitment without one, and that difference belongs in your thinking when a client treats a nearby branch as a durable feature of the lender they are choosing.

Who the moratorium does not cover

The moratorium arrangements run to the major banks. Canstar reports they do not cover non-majors, and that Bank of Queensland, Bendigo and Adelaide Bank, Heritage and People’s Choice each reduced regional branch numbers over the year.

For brokers, that creates a genuinely awkward crossover. Several of those lenders are the ones a regional or outer-regional client is most likely to name warmly — the local mutual, the bank the family has used for three decades. On these figures, that is also the part of the panel with no closure commitment attached to it. The sentiment and the shelter point in opposite directions.

Access to cash is thinning through the other channel too. Canstar’s count has Bank@Post outlets down 61 over the year to 3,304, with ANZ having become the last major bank to join the service in December 2025.

Why this is a broker question

Two weeks of industry coverage will file this as a banking story. It is a distribution story, and brokers are now the distribution.

The MFAA reported on 3 September 2026 that brokers facilitated a record 81.6% of all new residential home loans in the June 2026 quarter, representing $139.08 billion in new home lending. Set that beside a proprietary network that has lost 1,215 branches in five years and the shape of the market is hard to miss.

It is worth being precise about what that juxtaposition does and does not show. These are two separate releases, measuring two different things, and neither establishes that branch closures caused the broker share — or the reverse. What the pairing does establish is where the obligation now sits. If the branch is no longer the place a borrower gets advice about a mortgage, then the broker is the only party in the transaction with a statutory duty to act in that borrower’s best interests. The channel that carries 81.6% of new lending is the channel carrying the duty.

Where branch access meets RG 273

ASIC’s Regulatory Guide 273, Mortgage brokers: Best interests duty, issued 24 June 2020, is explicit that price is not the whole of the assessment. RG 273.57 states: “Cost is not the only matter relevant to whether recommending a product is in the consumer’s best interests.”

The guide also names the kind of preference a branch represents. Among the matters RG 273.48 lists for consideration is “the consumer’s priorities and preferences for different products or credit providers”, including the credit provider’s level of service. RG 273.71 goes further, noting that “some consumers may also derive value from other loan features” and giving two kinds of example: substantive features such as an offset account or redraw facility, and matters that are harder to quantify, such as the service levels and policies of the credit provider. RG 273.82 adds that judgment matters particularly where a consumer holds strong preferences or strongly held beliefs about certain credit providers.

So a client’s preference for a lender with a branch nearby is a legitimate, recognised, non-cost consideration. It is not a loophole and it is not a problem. The point is what RG 273.75 then asks of you: where a product is recommended on the strength of one of those features, you should be able to make an overall assessment that the non-cost consideration realistically offers that individual consumer greater value.

None of this is advice on your obligations, and it is not a statement of what any licensee requires. Record-keeping standards and the wording your file notes should carry are matters for your licensee or compliance adviser, and RG 273 should be read in full rather than through the paragraphs quoted here.

The cash conversation

Cash access is the version of this issue clients raise unprompted, usually on behalf of a parent or a cash-handling business. Australian Broker’s report cites a 2025 Reserve Bank survey finding that one-third of Australians would face hardship or major inconvenience if cash became hard to access — which is a useful figure to have when a client’s objection sounds like a minority concern and is not.

That is a servicing question rather than a credit question, and it is often better answered by mapping the client’s actual cash needs onto what the lender and Bank@Post still offer locally than by changing the lender on the application. Tindall noted Australian Banking Association work exploring options such as multi-bank hubs, while cautioning that any such model has to improve access without undermining competition.

What to review this week

What to watch next

Three things sit on the calendar. APRA’s next Points of Presence release is due in October 2027 and will be the first to measure a period running up against the NAB and ANZ expiry. Any extension, renegotiation or lapse of those commitments will be announced long before the data shows it — watch for it from mid-2027, or earlier if the arrangements are revisited. And the non-major reductions Canstar identified are the ones to track quarter by quarter, because they are the ones with no public commitment governing them.

The strategic read is simpler than the numbers. For twenty years, “which bank has a branch near you” was a question a mortgage conversation could reasonably lean on. On APRA’s own series, that question is becoming harder to answer durably, and for two of the four majors the most reliable part of the answer has a 2027 date attached. Brokers who have already moved the conversation from where the lender is to what the lender does will barely notice. For everyone else, this is a cheap week to update a sentence that has been sitting in client files for years.