Canadian Bank Earnings: What PCLs and CET1 Actually Tell You

The Big Six report on a late-August and late-October rhythm, and the market reaction is usually driven by a single line: provisions for credit losses. Most commentary treats PCLs as a measurement of loans going bad. Under IFRS 9 they are something quite different — a model-driven forecast — and that distinction is the whole analytical game.

How IFRS 9 provisioning works

Loans sit in one of three stages:

• Stage 1. Performing. The bank provisions for expected losses over the next twelve months.
• Stage 2. Significant increase in credit risk since origination, but not yet impaired. The bank must now provision for expected losses over the entire remaining life of the loan. This is a large step up.
• Stage 3. Credit-impaired. Provisions reflect actual expected loss on the specific exposure.

The critical consequence: the largest single driver of a provisioning surprise is migration between Stage 1 and Stage 2, not actual defaults. A loan that has not missed a payment can generate a substantial provision the moment the model reclassifies it, because the provisioning horizon jumps from one year to lifetime.

Banks disclose the staging table. It shows gross loan balances and allowances by stage. Reading it quarter over quarter answers the question the headline PCL number cannot: is credit deteriorating, or did the macroeconomic assumptions change?

Performing versus impaired provisions

Every Canadian bank splits PCLs into two components:

PCL on impaired loans reflects borrowers actually in trouble. It is backward-looking and hard to argue with.

PCL on performing loans is the model output — a function of forecast unemployment, GDP, housing prices and commodity prices, weighted across optimistic, base and pessimistic scenarios. Banks disclose the key macro variables and sometimes the scenario weights.

This split is where the interpretive work happens:

• Rising impaired PCLs with flat performing PCLs: real, present credit deterioration.
• Flat impaired PCLs with rising performing PCLs: the bank's economists got more cautious. Possibly prescient, possibly conservatism ahead of a difficult year.
• Falling performing PCLs boosting earnings: a release. It flatters EPS and tells you nothing good or bad about the loan book. Beats driven by releases deserve a discount.

Net interest margin, decomposed

NIM is net interest income divided by average earning assets. It moves for reasons that are worth separating:

• Deposit mix. A shift from non-interest-bearing chequing accounts to term deposits raises funding cost without any change in policy rates. This has been the quiet margin story through the high-rate period.
• Asset repricing. Mortgages reprice on renewal, so the loan book takes years to fully reflect a rate change. This produces a lagged tailwind or headwind independent of current policy.
• Wholesale funding costs. Sensitive to credit spreads and to domestic funding conditions.

Banks disclose NIM by segment. Canadian personal and commercial banking NIM is the cleanest read on domestic conditions; capital markets NIM is noise for this purpose.

CET1: the constraint on everything else

Common Equity Tier 1 capital divided by risk-weighted assets is the ratio that determines whether a bank can buy back stock, raise the dividend or make an acquisition. OSFI sets a minimum plus a domestic stability buffer, and the buffer level is itself adjusted by the regulator based on system-wide vulnerabilities.

Three things to check:

1. The reported ratio against the regulatory requirement. The gap is the discretionary capital.
2. Whether RWA growth is outpacing loan growth. Rising risk weights on a stable book mean the portfolio is being assessed as riskier — a signal in itself.
3. What the bank says about buybacks. A suspended or unrenewed normal course issuer bid is a capital statement, and it usually precedes the commentary.

A reading order for the quarter

1. Staging table — Stage 2 balances versus last quarter.
2. PCL split — impaired versus performing.
3. Macro assumptions in the provisioning note — did unemployment or housing forecasts change?
4. Canadian P&C NIM and deposit mix commentary.
5. CET1 versus requirement, and buyback status.
6. Only then, adjusted EPS.

Reversing that order — starting with adjusted EPS — is how investors end up owning a beat that was a provision release and selling a miss that was a conservative model update. The number the market reacts to in the first ten minutes is rarely the number that matters in twelve months.