2026-10-19 → 2026-10-21

Statistics Canada publishes September CPI at 14:30 CEST, 08:30 in Toronto, and the monthly headline is the line that carries the gasoline mechanics. Consensus is +0.2% m/m against -0.1% in August. The sign change is the whole forecast: August's -0.1% was a fall on an unadjusted basis even as the seasonally adjusted measure rose 0.2%, and the move back to positive is expected to come almost entirely from motor fuel, the same channel that took the American headline to 3.7%.

The annual rate is where the stakes sit. Canada has run 3.0% y/y in each of the last two months, July and August, and analysts expect September to tick up to roughly 3.2%. Excluding gasoline the August rate was 2.4%, up from 2.2% in July - a 60bp gap between the headline and the ex-energy measure that has been widening rather than closing. A +0.2% monthly print keeps the headline pinned near 3% and keeps the Bank of Canada's communication problem exactly where it is.

The hawkish case is straightforward and narrow: a print at +0.4% or above, particularly if the strength is outside fuel, would mark the first evidence that the energy pass-through has broadened in Canada the way it has in the United States. The Canadian dollar would firm, the front of the Canada curve would cheapen, and the 28 October decision - which comes with a Monetary Policy Report, the quarterly forecast round, and is already described as live - would be genuinely in play. The dovish case is a flat or negative print, which would read as August's softness repeating and would push the hike argument into December with no cost.

What makes this release less decisive than its billing is that the monthly headline is not the Bank's target variable. The Governing Council held at 2.25% on 2 September, its seventh consecutive hold, and said upside inflation risks had grown while declining to act on them. The two lines that bear on that judgement are the median and trimmed measures that print in the same minute, and both are expected unchanged. A hot m/m headline with unchanged cores is the configuration the Bank has already looked through twice.

Median CPI is the Bank of Canada's preferred core measure - the year-over-year change in the price component sitting at the 50th percentile of the basket, which discards the tails by construction rather than by judgement - and consensus has it unchanged at 2.0% y/y. Two percent is the target. A core measure printing exactly at target, in the same minute as a headline near 3%, is the cleanest statement available anywhere in the G10 of the distinction between an energy shock and an inflation problem.

That is why the forecast matters more than its lack of movement suggests. The Bank has held seven times running at 2.25% and justified each hold on the grounds that core is contained; the median measure is the evidence for that claim. An unchanged 2.0% renews the licence to hold on 28 October regardless of what the headline does, and the market has priced accordingly - odds of a quarter-point hike fell to roughly 34% after the soft September labour report.

The hawkish scenario is a median at 2.2% or above. It would be a small number with disproportionate consequences, because it would remove the Bank's stated reason for looking through the headline and would do so nine days before a decision that comes with a full forecast round. The Canadian dollar would rally hardest against the currencies whose central banks are further from acting, the two-year Canada would cheapen, and the 34% would not survive the afternoon. A print at 1.8% runs the other way and makes the October hold a formality rather than a judgement.

The reason to watch the median rather than the headline is asymmetry of information. Gasoline is observable in real time and already in the forecast; nobody is surprised by it. The percentile of the basket at which Canadian price-setting is actually running is not observable until this line prints, and it is the only number on the Canadian calendar this month capable of changing a committee's mind.

Trimmed CPI discards the most extreme 40% of the basket by weight - the top and bottom 20% of monthly price changes - and is expected unchanged at 1.9% y/y. Read alongside a median forecast at 2.0%, the pair says something the headline cannot: Canadian core inflation is not merely near target, it is near target under two different methods of throwing away the outliers. When the two measures agree to within 10bp, the tails are doing the work in the headline.

The joint reading is the useful one. Median and trimmed have both been stable - 2.0% and 1.9% in August - while the headline held at 3.0% and the ex-gasoline measure rose from 2.2% to 2.4%. The ex-gasoline series is drifting up and the trimmed measures are not, which locates the pressure in components volatile enough to be trimmed out but not narrow enough to be only fuel. That is the configuration that turns into a core problem if it persists, and the configuration that disappears if energy rolls over.

The scenario that matters is both core measures moving up together. Trimmed at 2.1% with median at 2.2% would be a 20-30bp shift on measures that have barely moved for months, and because they are constructed to be insensitive to outliers, a move of that size cannot be dismissed as a single component. That is the print that puts a hike on the table for 28 October. Both unchanged, which is the consensus, leaves the Bank where it has been since March and makes the Monetary Policy Report the event rather than the decision.

Trimmed CPI is the least likely of the three Canadian lines to surprise and the most informative if it does. It is also worth noting what this release is not: September data, published 19 October, nine days before a decision taken with the benefit of a business outlook survey released in the same window. The Bank will not be short of information on 28 October. It will be short of a reason to act, unless one of these two numbers supplies it.

The Office for National Statistics publishes September CPI at 08:00 CEST, 07:00 in London, and consensus is 3.4% y/y against 3.1% in August. August itself was an upside surprise, rising from 2.9% in July, and core held at 2.6%. A 3.4% print would be the third consecutive monthly increase in the annual rate and would take UK headline inflation to roughly twice the 2% target.

The Bank of England has already conceded the direction. Bank Rate was maintained at 3.75% in September, the Governor has written the open letter that a reading above 3% requires, and the Bank's own estimate as of 18 September puts inflation near 3.75% in the fourth quarter, attributing the rise to energy and motor fuel. So a 3.4% print is not news to the MPC - it is the path the MPC published. The information is in the gap between 3.4% and whatever prints, and in the core.

The hawkish scenario is 3.6% or above, or any print where core breaks out of the 2.6% range it has held since June. That combination would mean Britain is importing the American result - energy reaching the core - into a committee that is already above target, already writing letters, and meets on 5 November with a Monetary Policy Report. Sterling would rally, the gilt curve would cheapen at the front, and the November hold that is currently assumed would stop being assumed. A print at or below 3.2% does the reverse and reads as the Bank's forecast being too pessimistic, which is the better outcome for gilts and the worse one for sterling.

What makes this release genuinely two-sided is that the UK is the only one of this week's three jurisdictions where the policy rate is already restrictive by its own reckoning. Canada sits at 2.25% and New Zealand at 2.75%; Britain is at 3.75%, level with the bottom of the Fed's target range, with an economy whose August GDP was forecast to contract in last Thursday's release. A committee holding a restrictive rate into a rising headline and a flat core has a harder problem than one holding an accommodative rate into the same data, and more to lose from being wrong in either direction.

The 10-year gilt is the instrument that has already taken a view. It has traded in a narrow band around 5.39% through the first half of October and barely moved across the two published sessions of last week, which says the market expects this print to land near forecast and the MPC to do nothing on 5 November. A 3.4% print confirms that. Anything else is a repricing, and from a 5.39% starting level there is room in both directions.

Stats NZ publishes the September-quarter CPI at 23:45 CEST Wednesday, 10:45 Thursday morning in Wellington, and consensus is +0.8% q/q against +1.5% in the June quarter. New Zealand reports quarterly rather than monthly, which makes this the single most consequential inflation print on the global calendar this week: it is the last domestic data of substance before the Reserve Bank's Monetary Policy Review on 28 October, and there is no second observation to average it against.

The starting point is the worst of the three. Annual inflation reached 4.1% in the June quarter, up from 3.1% in March, the highest since the fourth quarter of 2023 and clear of the Reserve Bank's 1-3% target band. The Bank has responded: the OCR was lifted 25bp to 2.50% in July and again to 2.75% at the following meeting, two consecutive hikes after a period in which the Committee had split three-three and held on the chair's casting vote. New Zealand is the one economy this week whose central bank is unambiguously in a tightening cycle rather than deciding whether to start one.

The forecasts cluster tightly and that is what makes the release dangerous. The Reserve Bank's own expectation is +0.8% q/q, putting the annual rate near 3.9%; BNZ has 3.8%, Kiwibank 4.0%, Trading Economics 3.7%. A +0.8% print is therefore priced, and because the quarterly series is noisy the distribution around it is wide - a 0.3pp miss on a quarterly number is an ordinary event and would move the annual rate by the same amount.

Market pricing has already drifted. Traders put roughly a 58% chance on a hike to 3.00% on 28 October as of 6 October, down from about 80% in late September, with some houses arguing the Bank waits until December. A print at +1.0% or above, taking the annual rate above 4%, would push that back toward certainty: the kiwi would rally, the front of the New Zealand curve would cheapen sharply, and a fourth consecutive quarter outside the target band would make the Bank's framework, not just its rate, the subject of discussion. A print at +0.6% or below would do the opposite and hand the December camp the argument.

The reason to care about this outside New Zealand is the comparison it offers. Three central banks decide within nine days of each other on broadly the same inflation picture - headline above target, energy in the headline - and they sit at 2.25%, 2.75% and 3.75%. New Zealand is the furthest along the road the other two are still arguing about taking, and its print lands last. Whatever it says about whether tightening into an energy shock has worked is the most useful single data point this week offers.