Forbearance, staging and a breach of the large exposure limit

The Stage 1 conclusion is not defensible. Stage 1 applies only where there has been no significant increase in credit risk since origination. AASB 9 requires that assessment to weigh quantitative and qualitative indicators, not just payment status.

On the facts, four indicators point to SICR:

• Forbearance granted twice (extended interest capitalisation). By definition this is a concession the bank would not have offered but for the borrower's financial difficulty. The fact that payments are now current is the result of the concession, not evidence of health.
• Covenant breach on pre-sales, 40% against 65%.
• Collateral deterioration of 17 percentage points of LTV on a fresh valuation.
• Restructuring advisers appointed by the borrower.

Any one of these is sufficient on its own; together the exposure belongs in Stage 2, with lifetime expected credit loss and a PD reflecting forborne status rather than the unchanged grade 5 figure.

Stage 3 is a separate question and must be documented. The covenant breach and the distressed restructuring are unlikeliness-to-pay indicators. Whether they amount to default depends on the bank's default definition, but the assessment cannot be skipped simply because no amount is 90 days past due.

What follows: the exposure must be flagged and reported as forborne; lifetime ECL replaces 12-month ECL; the internal rating must be re-reviewed; and the RM's argument conflates arrears status with credit quality. A grade that cannot move on this evidence is a model risk finding, not just a credit judgement.

Single-name limit. Halcyon's connected exposure is A$200 million against Tier 1 capital of A$700 million - 28.6%, above the 25% large-exposure limit. The connected-counterparty test should aggregate the construction facility and the investment loan and consider related special purpose vehicles and guarantees. The limit breach requires Tier 1 capital growth, a syndication or participation of at least A$25 million, or a board-approved remediation plan with a timetable. Keeping the exposure at Stage 1 does nothing about the breach.

Sector concentration. Commercial property at 38% of the book and 210% of eligible capital is a one-factor bet on two cities. The IRB foundation formula is a single-factor asymptotic model and does not charge for sector concentration, so economic capital and stress testing must carry it, alongside APRA's IRB concentration expectations.

Stress design - apply the shocks simultaneously, not one at a time:

1. A 25-30% fall in commercial valuations, taking LTV from 79% to roughly 105-115%.
2. A 12-month freeze on pre-sales, which extends the construction facility and increases capitalised interest.
3. A 100bp cap rate rise, raising debt service and lowering the achievable sale price.
4. A second-round effect on other property exposures sharing the same collateral type and the same two precincts, since default correlation rises in exactly this scenario.

Report the outputs in the language the committee uses: lifetime ECL on Halcyon, the increase in credit risk-weighted assets, the IRB capital impact, and the impact on the single-name and sector limits. Then test whether the limit framework would have forced action before 38% was reached.

Why an unchanged PD is a red flag. An internal rating that does not react to forbearance, covenant breach, a 17-point LTV deterioration and the appointment of restructuring advisers is not measuring risk. Two possible causes, with different fixes:

• Deliberate through-the-cycle calibration, in which case the PD is stable by design but must be accompanied by a point-in-time assessment for provisioning. AASB 9 ECL needs a PD that responds to current conditions plus forward-looking information, whatever the IRB grade does.
• A stale or overridden rating, in which case this is a rating model failure and a model risk issue that likely affects the whole grade 5 pool.

Why it matters beyond Halcyon: if the rating system is slow to migrate, the bank's IRB capital, its provisioning, its watch-list population and its risk appetite reporting are all understated simultaneously. PDs that lag the cycle flatter capital exactly when capital is most needed, and rating migration is one of the first things a supervisor or an independent validator tests.

What I would require:

1. A documented rating re-review for Halcyon within 30 days, with the override rationale and approving authority recorded.
2. A back-test of the grade 5 pool - has any exposure with a similar profile migrated in the past? If not, the calibration is suspect.
3. A review of the bank's rating philosophy statement to confirm whether it is through-the-cycle or point-in-time and how the two interact with AASB 9 staging.
4. A validation finding raised in the model risk register, with board risk committee visibility, since this affects the reliability of capital and ECL simultaneously.