Australian major banks: Credit Quality — where the faultline runs (July update)

The shift is starting to materialise in their headline bad-debt numbers, particularly in regards to the direction that banks' own overlays have been trending, including any leading indicators across the non-bank lenders. 

Notably, prime consumer and mortgage credit is the one segment that has not been a meaningful strain on the banks' books yet. Historically, they tend to lag behind the already emerged stress in non-performing credit cards and/or asset financing segments. Therefore, we are anticipating for a broader consumer stress to appear as a 2H 2027 and FY 2028 risk. 

Our valuation models imply between 24% to 59% downside across the majors, and we view the principal determinant of whether losses track our base case or move toward our bear scenarios, on the basis of how the path of unemployment evolves. 

Lastly, we believe the market is underestimating the earnings and dividend risk associated, during these early stages of credit loss normalisation.

Provisioning across the majors

The provision normalisation we outlined in April (link to article here) has begun to appear in the banks' own numbers rather than only in our forecasts. 

The key observations from the latest results:

National Australia Bank (ASX:NAB) lifted from a higher base, consistent with its larger business-lending mix. FY25 cost of risk was 11bps (A$833m); the 1H26 run-rate had inflected to roughly 18bps annualised, write-offs doubled year-on-year, impaired assets rose 60%, and management added a A$300m Middle East provision overlay with a 45% weighting on the downside scenario. Our model peaks the charge at 17bps (A$1,369m) in FY26.

Commonwealth Bank (ASX:CBA) reported a benign first half struck on December data, with arrears falling and provisions released, but its May quarterly update showed an early turn: 90-day mortgage arrears rose 6bps to 0.71%, personal-loan arrears rose 30bps, corporate stressed exposures reached 0.94% of commitments, and management added a A$200m collective provision overlay. The move from releasing provisions in the first half to adding an overlay in the third quarter suggests management's own view of the outlook has shifted.

The FY25 charges sum to A$2,424m, and so far, the per-bank trajectories above are consistent with that path. Based on the data's readings, the ramp-up in provisioning has moved from being a forecast, to now being a reality, with three of the four banks having now increased their provisioning/coverage.

The odd-one out

Nevertheless, we maintain the view that the rate is at cyclical lows instead of being structurally lower - reinforced also by several of ANZ's own initiatives and metrics i.e. 1) took a A$175m Middle East overlay, 2) institutional collective-provision charge rose roughly sevenfold year-on-year (A$104m from A$15m), 3) and 90-day mortgage arrears reached 0.83%. 

With the combination of a low headline loss rate, rising arrears and a sharply higher institutional provision charge, this is more consistent with a lagging book rather than a durable improvement. Hence, we expect for ANZ to play catch up on normalising its loss rates to be in alignment with the sector.

Leading indicators from non-bank lenders

However, the specialty lenders which sit one step below them in the credit chain report sooner. Another reason why we believe it is important to track them. Across SME, asset-finance and non-conforming books, indications are consistent with an early-stage deterioration.

Paradoxically, its collective provision ratio has also fallen (as a share of loans) even while non-performers increased, suggesting that provisioning is potentially lagging the pre-emptive/actual losses that the book has already identified. 

If we assume the historical characteristic of SME-related credit quality deterioration leading retail's by around 6 to 9 months, or that it leads a rise in headline unemployment by around 9 to 12 months, the near-term outlook for both the majors, or the Australian economy, does not bode well.

Butn (ASX:BTN), an invoice financier distributed through SME accounting platforms, had already shown that receivables more than 150 days overdue, already more than doubled in a single half (to the start of the year), while write-offs still remained near zero - indicative that borrowers are paying more slowly instead of  defaulting for now. Patterns that typically precedes formal default by one to two quarters. Its funding cost rose to 67% of revenue from 56%, which illustrates the effect of higher rates on leveraged lenders.

Resimac's (ASX:RMC) data is bifurcated: prime mortgage arrears are improving (Stage 3 loans at 1.76%, from 2.11%), but vehicle and asset-finance cohort carried elevated write-offs of up to A$9.3m.

And for Liberty Financial (ASX:LFG), they reported a first-half impairment charge that was 49% below the prior-year rate. Meanwhile, its pool of past-due but not-yet-impaired loans sits elevated at 2.20% of the book. The only benefit I can think of, for temporarily suppressing one's impairment charges, would be for the sake of printing a better earnings number in the current period. 

Opening Judo's Pandora box

Judo's June trading update had taken the cost of risk to approximately 88bps, and delivered a 10% profit shortfall against a share price already down roughly 46% over the year. And Judging by the reaction in its share price, the update came as a huge shock to the market, with JDO shares dropping by ~40% on the day of. 

Where we stood at the time, emotions were very much more subdued. 

The outcome landed on the bear-case loss rate we had modelled for some time now, and was consistent with all the aforementioned leading indicators above. For us, that decline largely reflects the unwinding of a premium in its share price, especially one which our valuation just could not support.

With the stock now trading at around 90 cents, or about 0.6x tangible assets, the reaction may have overshot to the downside in the interim. Regardless, trading below our valuation (whose quality has decayed since its results release in February too) is not on its own an investment case. And against the backdrop of SME credit deterioration, we don't see sufficient fundamental support yet, and await the company's upcoming August results, before we are able to make better judgement on its investability. 

Prime consumer and mortgage credit

Plenti (ASX:PLT), which lends to prime consumers, indicates the same: 90-day arrears of 42bps on a high-quality book, a net loss rate of 0.94% that sits below its through-cycle level of around 1.2% and the consumer-industry average of approximately 1.5%, and origination growth of 32% written into a weak consumer-confidence environment, the risk from which typically emerges only after 12 to 18 months. 

On this evidence, prime consumer credit is the segment yet to turn.

The macroeconomic backdrop

On an interest rate setting, the RBA had cut to a 3.60% trough late last year and has since raised the cash rate three times, to 4.35%, a re-tightening into a slowing economy. NAB's business confidence index has fallen to -13.8, having reached -28.7 in March, a weak reading that maps onto NAB's business book. Consumer sentiment is around 80, indicating pessimism. Unemployment has risen to 4.4%, with an April reading of 4.5%, approaching but not yet reaching the 5.0% level at which our models shift from orderly normalisation to bear-case stress. 

The key culprit/driver behind the central bank's (incl. int'l peers) decision was spurred from the US-Israel/Iranian conflict, and a subsequent list in the price of oil/energy. 

While headline inflation has eased to 3.2%, the central bank is likely to maintain (or even raise) rates in the near-term, to address underlying/bubbling pressures across services inflation, or those derived from the fuel complex.  

On balance, the direction of borrower serviceability - i.e. for rates to be lowered, will remain unfavourable.

Implications for the index and passive flows

In June (link to article here) we estimated that the ASX 200 traded around 29% above its aggregate intrinsic value, with the overvaluation concentrated in Materials and Financials, together 63.7% of the index, and supported by passive flows that allocate by market-cap weight without reference to valuation. 

Commonwealth Bank alone represents 8.17% of the index, and our work implies roughly 60% downside to its fair value. Because passive flows are price-insensitive by mandate, they do not distinguish between a bank loss rate of 4bps and one of 18bps. 

In our view this creates a disconnect between the majors' elevated share prices and the credit trends now emerging in their loan books, and the concentration of those names in the index, combined with continued passive inflows, is a source of vulnerability should sentiment toward the sector shift.

Valuation and what we are monitoring

In our assessment, this is more a question of earnings and valuation rather than solvency.

Below approximately 4.5% we regard the cycle as manageable and our base-case fair values as appropriate; above 5.0% the loss-rate paths above move toward their bear cases and our fair values toward the lower end of their ranges. 

We would revisit our view on evidence that loss rates are stabilising rather than building, on unemployment holding below 4.5%, or on the RBA reversing course as the fuel shock fades. 

Absent those developments, we believe the market is pricing the majors for a credit environment that three of the four banks have themselves begun to move away from.

Conclusion

Bank capital is not in question, but we believe sector earnings and dividends carry more downside risk than current prices reflect. Our below-market valuations imply 24% to 59% downside, and we are monitoring unemployment and non-bank credit indicators to gauge the pace of any further deterioration.