2026-10-05 → 2026-10-09

This is a scheduled speaking engagement rather than a set-piece policy speech, and it should be treated as such. As of the weekend the Bank of Japan's own 2026 speeches index carried no entry for 6 October - the only October listing was Deputy Governor Uchida on the 5th, on AI, big data and monetary policy - so neither text nor venue was public ahead of time. What makes it worth watching is timing, not billing: it is Ueda's first appearance on this calendar since the Bank raised its policy rate on 18 September, and it falls 23 days before the Policy Board's next decision.

That September decision is the context. The Bank lifted its key short-term rate by 25bp to 1.25%, the highest since 1995, on a 7-2 vote, with Toichiro Asada and Ayano Sato dissenting in favour of a hold. The justification was pre-emptive: a risk that inflation deviates upward beyond the 2% target rather than evidence that it already has. The market's reaction was the interesting part. The yen weakened after a hike, because two dissents were read as a Board less committed to the next move than the decision itself implied - and it has kept weakening since, through 158 per dollar on 1 October and around 157.8 on the 2nd.

So Ueda arrives with a specific problem to answer, and two ways to answer it. The hawkish route is to point at the yen and at the 29-30 October meeting, which carries the quarterly Outlook Report and therefore the forecast revisions that normally precede a move. Language framing that meeting as live would lift the front of the JGB curve, where the 2-year sits near 1.92%, and give the yen the support that September's hike failed to deliver. The dovish route is to lean on the dissents - to say the Bank has time, that the 1.25% level already reflects the inflation risk, and that it will wait for the Outlook Report rather than pre-commit. That reads as tolerance for a weaker yen and would push the long end of the JGB curve, already at 4.148% in 30s, further out.

The pricing says the second route is expected. Roughly 83% implied odds of no change in October, against 24 of 66 surveyed economists looking for 1.50% at either the October or the December meeting - a market that believes the next hike is coming but not yet. In isolation, unscripted remarks three weeks before a decision rarely move that. What gives them more weight than usual is that the 10-year JGB has now cheapened for three of the last four weeks and closed last week at 3.097%, 1.5bp higher, while every other long end in this brief either rallied or barely moved. Japan's curve is pricing something, and the Governor is the only person scheduled to say what.

The minutes of the 15-16 September meeting land at 20:00 CEST, 14:00 in Washington. They are the account of the first Federal Reserve rate increase in more than three years: a unanimous 12-0 vote to lift the target range 25bp to 3.75-4.00%, with the statement calling inflation elevated and judging that the action would support a timelier return to the 2% goal. Chair Kevin Warsh declined to call it tightening, describing it instead as removing a dose of accommodation, and named three things that had changed since July - the economy strengthened, inflation did not slow, and geopolitical tensions intensified.

The difficulty is that two of those three have since reversed. The Summary of Economic Projections published alongside the decision put the median end-2026 funds rate at 4.1%, implying one more hike this year, with 12 of 19 participants at 4.125% and four above it. It also put median 2026 core PCE inflation at 3.4% and the median unemployment rate at 4.1%. August core PCE printed 3.0% and September unemployment printed 4.2%. The Committee's own central forecast was wrong in the dovish direction on inflation and the hawkish direction on the labour market, within three weeks, and the minutes cannot know it.

What the market will read them for, therefore, is not the decision but its margin. A 12-0 vote can record consensus or compromise, and the 2027 dots suggest the latter: eight participants projected at least two further increases, six saw one, and four expected cuts. The commentary since has split along the same line. John Williams has said there is no rush for further increases while still calling one more this year reasonable; Vice Chair Philip Jefferson has signalled no urgency; Michelle Bowman, who speaks Tuesday, would prefer no further hikes in 2026 at all. If the minutes show that several participants viewed September as finely balanced, the 17% October pricing is vindicated and the front end has room to extend last week's rally.

The hawkish scenario is the one that would actually move things, precisely because it is not priced. An account that dwells on inflation persistence, on oil pass-through with Brent at $102.25, or on a majority that regarded one more hike as close to committed, would force a repricing from 17% toward the 36% of a week ago. That hits the 2-year first: it would hand back the 9bp it gained last week and flatten 2s10s from the front, reversing the one clean trend in this market. The dovish scenario is less symmetric - it confirms what is already priced, so the front end gains less than the hawkish case would cost it.

In isolation the minutes deserve modest weight. They are three weeks stale, they are a committee's prose rather than a decision, and minutes rarely overturn pricing after the data has already spoken. Two things argue for more attention than that. This is the only scheduled US release of the week, so there is nothing else for the market to trade. And it is the last formal Committee document before 27-28 October, which means it is the last chance to learn how the Fed thought about a hike before it has to explain why it is not delivering the next one.

Statistics Canada's September Labour Force Survey is forecast to show employment up 9,000 after August's 41,700 decline. Forecasters are clustered at the low end and barely above zero - CIBC has 5,000 - which is to say the consensus is not for a recovery but for a stall. August was the reversal of several months of improvement, and its composition was broad rather than concentrated: business, building and other support services fell 20,000, public administration 8,800, natural resources 7,700 and utilities 5,600, offset only by a 22,000 gain in manufacturing. Youth employment fell 19,000 and core-age 16,000; Quebec lost 19,000 and Ontario 18,000.

The policy context makes this the most consequential release of the week outside the United States. The Bank of Canada held its overnight rate at 2.25% on 2 September, a seventh consecutive hold since the cut of October 2025, and its next announcement is 28 October. This Friday's survey is the last employment reading before that decision. The Bank's own language is the reason it matters: it acknowledged that labour market conditions had improved, with unemployment edging down to 6.4% in July, but added that demand for labour remains subdued and that indicators point to continued excess supply. One more negative print turns that hedge into a trend.

September is also the first month in which both sides of the tariff dispute were fully live. US Section 338 duties of 50% on $27.6bn of Canadian goods began collection on 22 August, and Canada matched them dollar for dollar effective 8 September. The Bank judged the inflation consequence muted, on the grounds that most of the counter-tariffs fall on intermediate inputs such as steel and on goods with domestic substitutes. It made no equivalent claim about employment, and the manufacturing gain in August predates the Canadian measures entirely. This is the first clean read on the labour side of that trade.

The scenarios are asymmetric in a way the forecast distribution understates. A second consecutive decline would make the excess-supply case difficult for the Governing Council to argue around three weeks later, and should bring 28 October into play for a cut - CAD softer, the front of the Canadian curve richer. A strong rebound, 50,000 or above, would make August look like a one-month aberration and cement the hold. The awkward outcome is the forecast itself: 9,000 jobs is statistically indistinguishable from zero in a survey with this one's sampling error, and a print near consensus resolves nothing while leaving the Bank exactly where it is.

The unemployment rate is forecast at 6.5%, up from 6.4%. Not all forecasters agree even on the direction - CIBC has the increase, others have it holding at 6.4% - and the level deserves context either way: at 6.4% it sits roughly 0.7pp below where it was a year ago, so this is a labour market that has deteriorated recently from a position that had been improving, not one in distress.

The reason to read this line alongside the employment change rather than instead of it is that the two can disagree, and did. August lost 41,700 jobs and the unemployment rate did not move at all, because the labour force shrank enough to absorb the loss. The United States produced the mirror image three days ago: 29,000 jobs added, which on its own is near-stagnation, and the rate still rose to 4.2% because the labour force grew faster. A rate that holds at 6.4% on a weak employment print tells you about participation; it does not tell you about demand.

For the Bank of Canada the rate is nonetheless the cleaner instrument, because excess supply is the specific thing it says it is measuring and the unemployment rate is the most direct measure of it. A print at 6.6% or above would be the single strongest argument available for a cut on 28 October, since it would mark a clear break from the July improvement the Bank cited. A hold at 6.4% alongside a soft employment change produces the genuinely ambiguous outcome, and the Bank's seven-hold record suggests how it resolves ambiguity.