On record before Sept 2: the Bank holds
Every forecaster says hold, so saying it earns nothing. Here is the part almost everyone has backwards: the market prices a zero percent chance of a cut and a 22.8% chance of a hike. The Bank of Canada is not poised to rescue the economy from tariffs — its next move, whenever it comes, is more likely up. Below: our headline call at 88%, five sub-predictions with real risk in them, and the pre-committed rule for what moves our recession odds. All published before the decision, all graded in public on Sept 2.
Why we are more confident than the market
The market's 22.8% hike probability is, in our view, too high — and the reason is a date almost nobody is factoring in.
Canada's retaliatory tariffs take effect on September 8. That is six days after the decision. A central bank staring at a known, dated, self-inflicted price shock six days out does not move rates first and find out second. It waits for the CPI print that measures it. This single fact is why we put a hold at 88% rather than the market's 77%.
The rest of the case is that both alternatives are already dead:
The cut is gone. Q2 GDP grew 0.8% — 3.3% annualized — against the Bank's own 2.5% forecast, and Q1 was revised up to +0.1%. You do not ease into an economy that just beat your forecast by 80 basis points. Housing investment rebounded 2.5%, per-capita GDP rose 1.0%, and the TSX finished August at record highs.
The hike is premature. Headline CPI is 3.0%, but that is gasoline; CPI-trim and CPI-median are running near 2% — at target. There is a warning sign in the ex-food-and-energy series, which has posted three straight seasonally adjusted 0.3% gains, and that is exactly what makes the market's 22.8% non-crazy. But three months of firm core is a reason to signal, not to act, when a tariff shock lands next week.
Both risks point in opposite directions and roughly cancel. That is the textbook definition of a hold.
The predictions, on record
The rate call is the free one. These are the ones with risk in them — specific enough to be wrong, published before the fact, graded on Sept 2 whatever happens.
| # | Prediction | Confidence |
|---|---|---|
| 1 | The policy rate stays at 2.25%. Ninth consecutive hold. | 88% |
| 2 | The statement names retaliation-driven inflation as a two-sided risk — explicitly framing tariffs as both a growth drag and a price pressure, rather than treating them as purely disinflationary. | 70% |
| 3 | No forward guidance toward easing. The statement contains no language a reasonable reader would take as signalling a cut at the next meeting. | 75% |
| 4 | The loonie moves less than 0.5% on the decision. Sept 2 is not the story this week — Sept 4 US payrolls is, and the CAD is trading on the Fed differential, not the BoC. | 75% |
| 5 | No published tariff scenario analysis. The Bank stays qualitative on trade-war outcomes rather than printing numbered scenarios. | 65% |
| 6 | The next move is a hike, not a cut — whenever it comes. Market pricing after Sept 2 continues to show higher odds of an increase than a decrease at the subsequent meeting. | 60% |
What each outcome would mean
The pre-committed rule
Published in advance so it executes without discretion on the day:
1. A hold with balanced-risk language → Canada recession odds unchanged at 35–40%.
2. A hold with explicit easing guidance → odds to 40–45%. Guidance toward a cut means the Bank sees damage the data hasn't printed yet.
3. Any rate change in either direction → odds to 45–50%, per the outcome table above.
And the correction rule we published on Aug 30 still binds: event-driven odds changes are capped at half the move until hard data confirms. A policy statement is not hard data. A CPI print is. So even a surprising Sept 2 gets half a move on the day, and the rest only if the Sept 15–16 CPI backs it up.