A hypothetical downturn reported in the Reserve Bank's October Financial Stability Review would dent bank capital, but the review found banks could absorb the losses and keep lending.

What would happen to Australia's banks if the economy took a serious turn for the worse? The Reserve Bank of Australia's (RBA) October 2026 Financial Stability Review, released on 1 October, offers one answer in the form of a stress test: a deliberately harsh, hypothetical scenario designed to probe how much damage the banking system could absorb.

The answer, in short, is that capital would fall but remain substantial. Banks' Common Equity Tier 1 (CET1) capital ratio, a core measure of their loss-absorbing buffer, stood at 12.4 per cent in June 2026. Under the stress scenario, it would fall to around 11.6 per cent. The review concluded that banks could absorb losses while continuing to extend credit.

It is worth being precise about the headline's word "withstood". Banks have not lived through this scenario. It is a modelled test, and the result describes how the system would be expected to cope, not something that has happened.

The scenario, shock by shock

The stress test combines three simultaneous shocks. Set against today's actual conditions, each is severe.

1. A 20 per cent fall in property prices. Both housing and commercial property prices fall 20 per cent. By comparison, the August minutes reported national housing prices about 1½ per cent below their March peak (at that time), while Governor Michele Bullock said on 18 September that prices remained about 50 per cent higher than in early 2020. The review separately estimated that a 20 per cent fall in housing prices would push "around 5 per cent of mortgages" into negative equity, up from below 1 per cent of borrowers today.

2. Unemployment rising to 6.3 per cent. Bullock cited an unemployment rate of 4.6 per cent on 29 September. The scenario therefore assumes a rise of 1.7 percentage points. The RBA's own forecast, published with the August minutes, had unemployment rising gradually to 4.8 per cent by end-2028.

3. A cash rate of 5.6 per cent. The cash rate target is currently 4.60 per cent, effective 30 September, after four increases in 2026 totalling 100 basis points. The scenario adds another full percentage point on top.

Taken together, the scenario resembles a sharp downturn arriving alongside further monetary tightening, a combination designed to test borrowers' repayment capacity and the value of the collateral backing loans at the same time.

What the result means

The CET1 ratio measures a bank's highest-quality capital relative to its risk-weighted assets. It is the buffer that absorbs losses before depositors or the wider system are affected. A fall of around 0.8 percentage points, from 12.4 to around 11.6 per cent, indicates that the hypothetical losses would erode some of that buffer, but leave most of it intact.

Just as important is the review's second finding: that banks could keep extending credit. In a downturn, a banking system that pulls back sharply on lending can deepen the slump. The test result suggests that, on its assumptions, the system would not be forced into that position.

The review points to several features of bank balance sheets that support the result:

  • Provisions already in place. Loss provisions are about 0.7 per cent of total credit.
  • Well-secured problem loans. As of June 2026, over 90 per cent of housing non-performing loans and over 50 per cent of business non-performing loans were well secured.
  • Limited high-risk lending. The share of high loan-to-valuation (LVR) lending has risen since the expansion of the Government's 5% Deposit Scheme but "remains low overall"; new investor lending has moderated from elevated levels.
  • Household cash flow. Around 2 per cent of variable-rate owner-occupiers have a cash-flow shortfall, and most of those have savings to cover it for at least six months.

The review's media release summed up the system as having "a good degree of resilience, but global and operational vulnerabilities continue to mount".

What the test does not cover

A stress test is only as broad as its scenario, and the review itself flagged risks that sit outside a conventional downturn. It named geopolitical tensions, global financial market vulnerabilities, advances in AI and disruptions at critical service providers as threats. It warned that frontier AI is increasing the speed, scale and sophistication of cyber threats, and that "Cyber-attacks have the potential to cause disruptions at a systemic level", including through common service providers.

Beyond the banks, non-banks make up about 6 per cent of financial system assets, and the review said their risk is "contained by their relatively small size". Private credit accounts for about 10 per cent of total business debt but less than 2 per cent of financial system assets, though some private credit funds have had to enforce redemption limits.

In a separate exercise, an APRA stress test of superannuation found that switching to cash at four times the volume seen in the quarter at the start of the COVID pandemic could be largely funded from international equity portfolios, with all participating funds maintaining adequate liquidity.

How close is reality to the scenario?

On the RBA's current assessment, not very. Bullock said on 29 September that a recession is "not our central base case at this moment", and the Board's statement said labour market conditions had eased broadly as expected. Company insolvencies are "around their longer-run average", though elevated in hospitality and construction.

That does not make the scenario irrelevant. Its purpose is to test resilience against outcomes well beyond the central forecast, not to predict them.

The bottom line

In a hypothetical world of 20 per cent property price falls, 6.3 per cent unemployment and a 5.6 per cent cash rate, the October Financial Stability Review found banks' CET1 ratio would fall from 12.4 per cent to around 11.6 per cent, and that banks could absorb losses while continuing to lend. Existing provisions and well-secured problem loans support that result. The test is a measure of buffers, not a forecast, and the review was equally clear that cyber, geopolitical and operational risks continue to build outside its frame.