Rate Divergence

State (current)

The developed world is not in a single monetary regime, and the week of 17 August makes the split visible in one calendar. Almost every G10 central bank is HOLDING while explicitly describing inflation as too high and keeping a HIKE, not a cut, as the live option. The Fed held at 3.50-3.75% in July with three dissents from Logan, Hammack and Kashkari, all of whom voted to raise. The RBA held at 4.35% unanimously, said inflation is not expected to return to around the midpoint of target until late 2027, and Governor Bullock confirmed the Board did not discuss a cut, only whether to raise or maintain. Norges held at 4.25% on 14 August with guidance that it "may still become necessary to raise the policy rate", and SEB still calls for a final hike to 4.50% in September. The Riksbank is expected to hold at 1.75% on 20 August, a view shared by all economists surveyed by Bloomberg. The BoC is on hold with markets pricing roughly 13bp of tightening by year-end, about a 52% probability of one 25bp hike, and with the policy rate already near the lower end of its estimated neutral range.

The BoJ is the outlier and it is moving the other way. Money markets price around a 62% chance of a September hike, sources report the Bank is considering ACCELERATING subsequent hikes, and the July Summary of Opinions carried members arguing that the pace could exceed market expectations and that the cost of delaying is not insignificant. Japanese core CPI is expected at 1.8% y/y for July from 1.6%, with headline expected at 2.0%, which would return national inflation to the BoJ's target for the first time in recent months, driven mainly by IMPORT prices.

That last clause is the theme's organising insight. The Fed cluster is holding because domestic demand is cooling, and the US consumer is doing the cooling. The BoJ is tightening because imported energy is doing the heating, with Japan's June current account already having swung to a Y923bn deficit against a projected Y1,512bn surplus. Same global shock, opposite policy response, because one economy imports the shock and the others largely absorb it through the household. For an allocator this is the mechanism that makes currency and duration exposure non-trivial in a portfolio that would otherwise just own global beta.

The second live element is that the Fed's hold is not the same instrument it used to be. Chair Warsh has withdrawn explicit forward guidance, which makes every meeting effectively live, and the July decision produced a pronounced curve STEEPENING with the front end richening and the long end cheapening on higher term premium. Warsh signalled comfort that tightening financial conditions is doing some of the Fed's work, and there is open speculation about leaning on balance-sheet policy alongside the funds rate. A hold that is delivered through the long end and the balance sheet is not the same monetary stance as a hold delivered through the policy rate alone.

UPDATE 18 August: the theme's first test came back against its Japan leg and in favour of its organising insight. Japan's Q2 GDP grew 0.3% q/q and 1.1% annualised against consensus of 0.5% and 2.0%, and the composition is worse than the headline. Private consumption printed -0.0% q/q, its first negative quarter in eight, capital expenditure fell 1.2%, domestic demand subtracted 0.2 percentage points, and the entire growth contribution came from external demand at +0.5pp, helped by falling imports as much as rising exports. Real employee compensation rose 0.8-0.9% q/q, so Japanese households are getting paid more and spending less. Sources framed it directly as complicating the BoJ's hike timeline. The September hike was priced around 62% before this print, and Friday's core CPI, expected at 1.8% from 1.6% on import prices, is now carrying more weight than it was.

That result sharpens rather than breaks the theme, because it makes the cost-versus-demand split explicit inside a single economy. Japan's inflation is arriving through the import bill while its domestic demand contracts, which means the BoJ is being asked to tighten into a consumption recession caused partly by the same energy bill that is producing the inflation. That is a harder policy problem than the one the G10 hold cluster faces, and it is the reason a September BoJ hike was never a clean read.

Canada supplied the theme's other new data point and it is the more useful one. July CPI accelerated to 3.0% y/y, the ceiling of the BoC's 1%-3% band, with gasoline up 25.7% y/y. Strip that one line and CPI ex-gasoline sat at 2.2% for a third consecutive month, with CPI-median at 2.0% and CPI-trim at 1.9%. Economists quoted said the rebound is not enough to sway the Bank, whose next decision is 2 September. That is the cleanest published evidence available anywhere in the G10 on the question every central bank in this theme is actually asking: does the energy shock reach core, or does it sit in the headline and pass through. In Canada in July it sat in the headline. One country, one month, and it is evidence rather than proof, but it is the kind of evidence that should move a view.

Timeline

• 2026-08-18: First test of the theme, split result. JAPAN Q2 GDP prelim +0.3% q/q and +1.1% annualised vs cons +0.5% and +2.0%, a deceleration from Q1's +0.5%; PRIVATE CONSUMPTION -0.0% q/q, FIRST NEGATIVE IN EIGHT QUARTERS (cons +0.5%); capex -1.2%; domestic demand -0.2pp; external demand +0.5pp the primary and only driver, supported by rising exports AND falling imports; real employee compensation +0.8-0.9% q/q. Sources: undershoots forecasts, COMPLICATING the BoJ hike timeline; underscores weak domestic demand. BoJ September hike was ~62% priced pre-print; Japan July CPI due Friday, core expected 1.8% y/y from 1.6% on import prices. CANADA July CPI 3.0% y/y from 2.8%, at the CEILING of the 1-3% band, +0.5% m/m and +0.3% SA m/m, GASOLINE +25.7% y/y, CPI EX-GASOLINE +2.2% for a THIRD consecutive month, groceries +3.1%, shelter +1.3%, CPI-median 2.0% and CPI-trim 1.9% (average 2.0%); headline beat expectations while underlying stayed contained; economists say not enough to sway the BoC; decision 2 Sept; ~13bp priced by year-end. CHINA July activity all missed: industrial production 4.5% y/y (4.8% exp, 5.3% prior), retail sales 0.6% (1.5% exp per Reuters poll, 1.0% prior), urban FAI -6.7% Jan-Jul (-6.0% exp, -5.7% at H1); PBoC LPR decision Thursday, hold expected at 3.00%/3.50%. US side: September hike odds fell to ~30-31% while the 30Y hit 5.31%, highest since June 2007 (https://investinglive.com/news/japan-q2-gdp-growth-undershoots-forecasts-complicating-boj-s-hike-timeline/ ; https://finance.biggo.com/news/cdfb8151-063f-4621-af72-a320b778dc47 ; https://www150.statcan.gc.ca/n1/daily-quotidien/260817/dq260817a-eng.htm ; https://www.rbc.com/en/economics/canadian-analysis/data-flashes/canadian-inflation-edges-higher-in-july-while-underlying-pressures-remain-contained/ ; https://www.cnbc.com/2026/08/17/china-economy-sales-investment-july-.html ; https://www.cnbc.com/2026/08/17/treasury-yields-federal-reserve-fomc-minutes.html)
• 2026-08-17: Entity created. The week's calendar puts the divergence in one frame. Fed held July at 3.50-3.75% with three hike dissents (Logan, Hammack, Kashkari), no forward guidance, September hold priced around 67% as at 15 Aug; minutes Wed 19 Aug. RBA held 4.35% unanimously, inflation not back to target midpoint until late 2027, no cut discussed, three increases already this year. Norges held 4.25% on 14 Aug with core CPI at 2.7% y/y for June and July, below its own 3.3% forecast, yet kept hike guidance; SEB expects 4.50% in September. Riksbank expected to hold 1.75% on 20 Aug, unanimous among economists surveyed by Bloomberg, no new forecasts. BoC on hold with ~13bp priced by year-end (~52% chance of a 25bp hike), prior CPI 2.8% headline and 2.1% core, preferred-measure average 2.1% from 2.27%; July CPI due today. BoJ the outlier: ~62% priced for a September hike, sources say it is considering accelerating subsequent hikes, July Summary of Opinions hawkish, core CPI expected 1.8% y/y for July (from 1.6%) and headline 2.0%, driven by import prices. PBoC expected to hold LPR at 3.00% (1yr) and 3.50% (5yr) on 20 Aug, favouring targeted liquidity and fiscal support over broad easing, with bank net interest margins limiting cut scope (https://global-view.com/fomc-newsquawk-week-ahead-in-focus-17th-21st-august-2026/ ; https://www.fxstreet.com/news/japan-gdp-steady-inflation-near-boj-target-ing-202608141420)
• 2026-08-16 (context, already banked LQB-0072): Australian rates market pricing 3bp for September and 13bp of hikes by year-end; NAB business confidence steady at -6. Japan June current account swung to a Y923bn deficit vs a projected Y1,512bn surplus (https://www.ig.com/en-ch/news-and-trade-ideas/week-ahead--17-august-2026-260814)

Our calls (journaled)

• 2026-08-17: The useful read is not "who hikes next" but WHY each central bank is where it is, because the reason determines what a portfolio should own. The G10 hold cluster is a demand story: inflation is above target, growth is cooling, and the household is the cooling agent, which is why the US consumer print on 14 August moved global rate expectations more than any central bank statement did. The BoJ is a cost story: inflation is arriving through the import bill while the current account has already swung to deficit, which is why an energy shock that is disinflationary for the US through demand destruction is inflationary for Japan through the terms of trade. Two testable implications this week. First, if Japan's Q2 GDP holds near 0.5% q/q with private consumption carrying it, and Friday's core CPI arrives at or above 1.8%, the September BoJ hike becomes the most likely single policy move in the developed world this quarter, and that is a yen and JGB event before it is an equity event. Second, if FOMC minutes show the three dissenters had broader sympathy, the divergence NARROWS rather than widens, because the market is currently pricing the Fed's first hike into 2027 and the committee may not be. Invalidation of the divergence framing: a Japanese GDP miss plus a soft Japanese CPI that pushes the September BoJ hike out, which would collapse the theme into a single global hold. Watch the JGB curve and the yen, not the Nikkei. Outcome: open

• 2026-08-18: Score yesterday honestly, because this one went against us on the leg we named. We wrote two testable implications and stated an invalidation: "a Japanese GDP miss plus a soft Japanese CPI that pushes the September BoJ hike out, which would collapse the theme into a single global hold." The first half of that invalidation FIRED within a day. Japan missed on the headline and missed worse on the composition, with private consumption negative for the first time in eight quarters and capex down 1.2%. We also wrote that if GDP held near 0.5% with private consumption carrying it, the September BoJ hike would become the most likely policy move in the developed world this quarter. Consumption did not carry it, external demand did, so that conditional did not trigger. Call this a MISS on the Japan leg and log it as one, not as a framework note.

What survives, and why we are not retiring the theme: the miss made the ORGANISING insight more true rather than less. The claim was that Japan's inflation is a cost story arriving through the import bill while the G10 hold cluster faces a demand story. Japan just printed contracting domestic demand alongside inflation expected to reach target on import prices, with growth coming from falling imports. That is the cost story with the demand side removed, which is a harder version of the same thing. The BoJ is being asked to tighten into a consumption contraction driven partly by the bill that is causing the inflation. So the divergence is intact but its Japan leg is now less likely to resolve through a clean September hike and more likely to resolve through a delayed and defensive one. Friday's core CPI is the remaining leg of our own stated invalidation and we should treat it as live: if it comes in soft, we said the theme collapses into a single global hold, and we should honour that rather than renegotiate it.

The new and better-supported claim is the Canadian one, and we are stating it as this run's call because it is testable elsewhere. Canada's July CPI is the cleanest natural experiment in the G10 on whether this energy shock reaches core: headline 3.0% with gasoline +25.7%, ex-gasoline 2.2% steady for three months, core measures 1.9%-2.0%. If the shock is passing through the headline and not embedding, then the entire hawkish-hold cluster is holding for a reason that will look correct in hindsight, and the three Fed dissenters' argument that supply shocks plus AI capex could ENTRENCH inflation is the thing being tested rather than assumed. Testable both ways and dated: US core PCE on 26 August, Japan CPI Friday, UK CPI Wednesday. If two or more of those show energy in the headline but not the core, the divergence is about the SHAPE of each economy's energy exposure rather than about differing inflation risk, and duration in the hold-cluster currencies is less risky than the 30-year at 5.31% implies. If instead core measures start rising alongside the headline, the dissenters were early rather than wrong and the front end is mispriced. Watch UK CPI Wednesday and Japan CPI Friday. Outcome: open

Hit-rate

• 0 hits / 1 miss (1 closed). The 08-17 Japan call missed within one day: we made the September BoJ hike conditional on GDP holding near 0.5% with private consumption carrying it, and Japan printed 0.3% with consumption negative for the first time in eight quarters. The stated invalidation for the whole theme is half-fired; Friday's Japanese CPI decides the other half. Lesson consistent with the bank-wide pattern: we again attached a directional conditional to the right variable. Japanese consumption WAS the right thing to name.