Are the banks well-provisioned? Or are they playing catch up?

The Australian banking sector approaches the May 2026 reporting season facing a structural turning point in the credit cycle - and a significant valuation paradox. The market continues to price the Big Four at 30–55% premiums to our fair value estimates. While low provision coverage has been fairly absent from consensus commentary, it has been a high-risk flag in our assessment for several months now. Only with the deteriorating macroeconomic environment are we starting to see banks forced to reassess their risk functions, through recalibration of their balance sheets.

The Catalyst: NAB and Westpac Pre-Announcements

The narrative for the May reporting season was set in mid-April when Westpac and NAB independently pre-announced significant provisioning lifts. Both banks cited the same macroeconomic shock - Middle East conflict, elevated fuel and energy costs, and their transmission throughout business lending books.

NAB announced a total 1H26 credit impairment charge of A$706m. The most consequential data point is the A$300m increase in forward-looking collective provisions, comprising a A$152m economic adjustment and a A$201m forward-looking adjustment for fuel-exposed sectors. Under IFRS 9's probability-weighted economic forecast settings, NAB has formally increased the downside scenario weighting by 2.5 percentage points to 45% - signifying a step change in its attitude towards future credit provisioning.

Westpac moved earlier, but with a softer signal, flagging a credit impairment charge of 10 bps of average gross loans (~A$445m) - versus 5.5 bps in FY25. The bank also raised its Collectively Assessed Provisions-to-Credit RWA ratio to 129 basis points and added an unquantified portfolio overlay for energy-intensive sectors.

To be clear, NAB and Westpac are not the only banks implementing defensive measures around loan book credit quality. What is more notable is the growing momentum behind risk-averseness across the sector. And that is a natural segue into our next point: an anticipated ramp-up in credit provisions across the board.

The Sector-Wide Provisioning Ramp

Provision coverage ratios across the banks have remained largely unchanged since the COVID era — an environment marked by low interest rates and forgiving anti-insolvency measures such as tax deferrals and fiscal handouts.

That goldilocks period does not last forever.

One of the key assumptions embedded in each of our bank valuation models is the normalisation of provision coverage. We are forecasting loss rates to trend from the current ~5–10 bps towards through-the-cycle historical averages of 15–20 bps.

Our projections show the total provisions balance held by the Big Four rising from approximately A$2.4bn in FY25 to a peak of roughly A$5.5bn by FY29. The ~A$3bn in cumulative provision additions will be a meaningful drag on cash earnings.

For example, CBA's loan loss rate is modelled to move from 7 basis points in FY25 to 20 basis points by FY29, pushing provision expense from A$726m to A$1.67bn. ANZ's trajectory moves from A$441m to a steady state of A$1.35bn–A$1.50bn by FY28.

Signals That Validate the Provision Ramp

Alpha Insights has been tracking the banks' asset quality for some time now, uncovering, collecting, and aggregating signals from across the market spectrum to pinpoint where in the cycle we might be.

Peers are normally a great place to start. And by all accounts, SME-focused Judo Capital (ASX: JDO) represents the cleanest direct indicator for NAB's heavier business banking exposure.

So, what has Judo been telling us?

JDO's 30-day past due loans have risen steadily from 2.5% in June 2024 to 3.94% in December 2025, accompanied by NPL ratios rising for five consecutive periods. This validates our outlook that credit quality is deteriorating and that provision coverage needs to increase.

Interestingly, Judo also reported record originations alongside rising arrears - which could suggest a late-cycle dynamic, where businesses draw down on available facilities to bridge cash flows instead of reducing leverage.

SME credit quality has historically led retail credit quality by 6–9 months, and headline unemployment by 9–12 months. This is an important lead-lag relationship for what follows.

At the smaller end of the market, we have Butn (ASX: BTN) - an invoice financing platform that facilitates payments and provides cash flow financing solutions for B2B clients. At their December results, the company noted that 150-day receivables had grown from A$1.5m to A$3.1m (+111% in 2H25 versus 1H25), while write-offs remained near zero. The extension in payment terms alone could represent a preliminary stress signal for the SME space, and an eventual pick-up in corporate defaults.

Another point of validation for our higher provision balance forecast.

The Agricultural Vice

NAB carries the largest agricultural loan book amongst its peers, A$43bn in absolute terms, and approximately 39% of its total SME book.

So, what can a company like GrainCorp (ASX: GNC) tell us about the health of NAB's agribusiness exposure?

GrainCorp's preliminary FY26 guidance is for grain receival volumes of 11–12 million tonnes, a decline of ~7% on FY25. In their guidance update, they stated that global grain markets continue to be characterised by cyclical oversupply and low prices. Importantly, this guidance was issued prior to the Iran conflict and the fertiliser supply disruptions now emanating throughout the industry.

It is not so much GrainCorp itself that we are interested in. Rather, it is the clients contributing to GNC's receival volumes - the farmers, who matter. The same farmers utilising GrainCorp's infrastructure are also likely to transact with NAB's agribusiness division.

If production volumes continue to fall, commodity pricing remains subdued, diesel costs climb, and two RBA rate hikes since February compound on top, the creditworthiness of those farmers has likely deteriorated. Which means their existing facilities with banks like NAB may be under-priced for risk.

To illustrate the monitoring framework: if GrainCorp were to issue a downgrade tomorrow, revising FY26 receival guidance from 11–12mt down to, say, 10–11mt or even 7–8mt, the downgrade itself would be a signal, and the magnitude of the revision would be indicative of conditions across the broader industry. All of that data feeds back into our intelligence on NAB's agribusiness exposure. 

This is one of many examples of how Alpha Insights builds inter-company relationships into a monitoring system that leverages the informational network across the ASX.

Mortgage Stress Is Still to Come

The provisions announced so far are almost entirely for business and SME lending. Very little has been discussed about consumer mortgage stress.

Helia Group (ASX: HLI) - our primary 12–18 month leading indicator for mortgage stress, reported a delinquency rate of 0.79%, a historical trough against a through-cycle normal of 1.5–2.5%. Helia's trough data implies that the current provision builds are entirely business lending in nature.

If we assume a transmission duration of approximately 18 months from trough to mid-cycle, we could expect mortgage-related credit impairment charges to inflect materially in 2H27 and into FY28. I should note that HLI's data is somewhat muddled by the internalisation of LMI operations across the major banks over recent years, effectively placing Helia in a managed run-off. Nevertheless, the directional signal remains useful.

Consumer discretionary data across the board confirms the squeeze is deepening. Domino's (ASX: DMP) has reported ANZ same-store sales of negative 4.7% across three consecutive halves. Step One's (ASX: STP) Australian revenue has dropped 35.6%. Economic stress is spreading through the system, casting a concerning outlook on both consumers' and consumer-facing businesses' ability to service loan obligations.

The AI-Employment Feedback Loop

There is an emerging structural pressure on the unemployment number that the market is not yet pricing. CBA announced a further 119 job cuts this week, on top of 300 roles already eliminated in February, with the Finance Sector Union flagging automation as a contributing factor.

This follows a broader pattern across corporate Australia. WiseTech (ASX: WTC) has flagged 2,000 role reductions over two years, Atlassian cut 1,600 positions, and Australian tech firms have shed approximately 4,450 workers in 2026 alone, many citing AI as a driver.

This dynamic warrants close attention. AI-driven efficiency gains are compressing headcount at precisely the point in the cycle where the labour market can least afford it. If banks themselves are cutting roles to fund technology investment while simultaneously building provisions for a weakening economy, the feedback loop between cost-out programmes and credit deterioration accelerates.

What's Next

The single most critical variable across our entire coverage universe is the unemployment rate, currently at 4.3%. Every model identifies 5.0–5.5% as the threshold that shifts provisioning from orderly normalisation to bear-case stress. Below 4.5%, this cycle is painful but manageable. Above 5.0%, our fair values move to the bottom of our ranges - CBA at A$60.00, NAB at A$20.38, ANZ at A$14.69, and WBC at A$21.50.

Bottom Line

The Australian banking sector has entered a structural margin compression regime. Rising credit provisions, upstream SME stress, and the emerging AI-driven employment headwind converge to create a material threat to earnings and dividends over the next 2–3 years.

At current prices, the market appears to be pricing the Big Four for sustained double-digit yield growth, pristine credit environments indefinitely, and zero probability of macroeconomic downside.

Not a lot of margin for safety.

Our fair values for CBA at A$77.00, NAB at A$26.44, WBC at A$27.16, and ANZ at A$21.40 - suggest that further retracement is required before the risk-reward starts making sense again.