The 3 per cent buffer survived. Who argued what, and what it actually costs you

The consultation on APRA's serviceability buffer closed on 18 July, and the outcome landed the way the smart money expected: the buffer stays at 3 percentage points. Every home loan in the country will keep being assessed at the loan's rate plus 3, which at today's typical 6.10 per cent variable means the bank tests whether you can service the debt at 9.10. The buffer has been the industry's favourite villain since it went to 3 in October 2021, when the cash rate was 0.10 and testing at 5-and-change felt conservative. At a 4.35 cash rate the same setting tests borrowers above 9 per cent, and the entire mortgage industry queued up this winter to say so. It did not work, and the reasons it did not work are worth understanding, because they tell you exactly what the regulator is afraid of and exactly where your file actually has room to move.

What the submissions argued

• The MFAA, the mortgage broker peak body, pushed a dynamic buffer: one that shrinks as rates rise and grows as they fall, on the logic that a fixed 3 points at the top of a cycle double-counts the tightening the RBA has already done. Elegant on paper, and precisely the kind of pro-cyclical dial APRA hates handing to a future version of itself under political pressure.
• The FBAA went with the big number: cutting the buffer by half a percentage point would release roughly $276 billion in aggregate borrowing capacity. That figure was supposed to be the headline argument for relief. Inside APRA it reads as the case for the defence: $276 billion of extra debt capacity, released into a market where Sydney and Melbourne prices are already correcting, is the risk, not the prize.
• The bank lobby argued for flexibility at the margins, particularly for refinancers trapped on high rates who fail serviceability at their own current repayment level, the mortgage prisoner problem. This is the one argument that has already partially won in practice: most majors quietly run dollar-for-dollar refinance exceptions at a reduced buffer for clean-conduct borrowers, and APRA has tolerated it.
• Consumer groups argued the buffer is the only thing standing between marginal borrowers and the arrears queue, and that the borrowers priced out by the buffer at 9.10 are exactly the borrowers who should not be writing cheques their income cannot cash at 9.10.

Why APRA held the line

Three reasons, none of them mysterious. Australian household debt-to-income sits among the highest in the developed world, so the system-level margin for error is thin. Arrears are still low but have been drifting up through the hike cycle, which is precisely when a prudential regulator refuses to loosen underwriting. And the buffer's bite is at its maximum exactly when the risk is at its maximum: the cycle top is when the marginal approved borrower is most likely to meet a recession with no headroom. APRA's institutional memory is 2019, when it dropped the old 7 per cent floor rate and watched borrowing capacity and prices rip. It is not making the same move at the top of a hiking cycle with a correction already running in the two biggest cities.

The worked numbers at 4.35

A couple on a combined $180,000 gross with no children, no other debts and ordinary living expenses, borrowing at 6.10 per cent variable over 30 years, gets assessed at 9.10. Under that test their capacity lands around the mid $700,000s at most mainstream lenders. Run the identical file with a 2.5 buffer, the pre-2021 setting, and capacity rises by roughly 5 to 6 per cent, call it $40,000 more. The FBAA's half-point argument, made personal. Notice what that does not say: it does not say the buffer is the binding constraint on your file. For a large share of applications the thing that actually caps the number is not the buffer at all, it is the expense benchmark the lender applies, the debt-to-income cap, or a HECS balance quietly shredding net income. Two lenders with the same 3-point buffer can be $80,000 apart on the same couple because their expense treatment, DTI ceilings and negative gearing add-backs differ. The buffer is uniform by regulation. Everything around it is not, and that non-uniformity is where files get won.

What changes from here

Nothing this quarter, which is the point. The practical consequences: pre-approvals keep being assessed above 9 per cent while the cash rate holds at 4.35, so borrowing power stays compressed into the spring market, which is quietly good news for buyers who already have their approval, because the competition is capped too. Refinancers who fail the full test should ask directly about dollar-for-dollar refinance exceptions, which live on despite the formal buffer staying put; the majors do not advertise them, brokers use them weekly. And when the rate cycle finally turns down, remember the mechanism runs in reverse: every 25 point cut adds roughly 2.5 per cent to capacity across the entire market simultaneously, which is how corrections end. The buffer review that actually matters will be the one APRA runs after the first cut, and on this week's evidence they will take their time with that one too.

Disclosure: Your Finance Guide partners with Australian Lending and Investment Centre (ALG) ACL 505575 for broker matching, and ALG receives lender commissions on settled loans. Worth saying plainly: brokers as an industry lobbied for a smaller buffer because smaller buffers mean bigger loans and bigger commissions. We think APRA got this one right anyway. The play is not borrowing the theoretical maximum, it is finding the lender whose honest assessment of your file is the least wrong.