The dollar no longer cashes in on inflation

Gasoline accounts for more than half of the rise in September CPI and for more than 70% of the rise in retail sales; the control group grew by only 0.1%, against +0.3% expected. Michigan sentiment fell to 46.3, described as its lowest in at least fifty years, and average hourly earnings are up 3.02% year on year, the lowest in the series in our dataset. Our reading is that traders saw in these figures prices rising on stagnating volumes.

The Fed endorsed that reading. After the CPI, the probability of a hike on 28/10 climbed to 43.7%, then fell back towards 27-33% after remarks from Williams and Daly; Warsh then declined to "prejudge" the meeting. The Daily of 12/10 judged that a +0.3% core print would reopen October: core did come in at +0.3%, the door opened for half a session, then the Fed closed it.

The curve flattened by 9 bp, to 35 bp, through the long end. The 10-year real yield moves from a 2.93% September average to 2.74% in October, with the breakeven stable around 2.35%: investors bought back duration without revising their inflation expectations. The sell-off in AI stocks contributed, with the Nasdaq down 4.5% on the week according to a single source and the VIX at 22.

The regime has changed since September: the dollar used to rise with the 2-year yield; it is now capped by a Fed that promises hikes without setting a date. The closing high of 07/10, at 102.41, would only come back within reach if the 2-year returned towards its 4.92% peak of 28/09, and between now and 28/10 the US calendar holds only jobless claims, housing data, the flash PMIs and the final Michigan survey.

The IEA and OPEC published opposing diagnoses of demand on the same day. The IEA expects a decline of 2.6 mb/d in 2026, the first since the pandemic, while OPEC still projects an increase of 0.2 mb/d, a gap of 2.8 mb/d. Crude buyers paid no attention: the Brent contract ended Friday at $109.86 and touched $110 intraday, a first since May, after at least nine attacks in the strait since early October and a missile fired near a major Saudi oil port. According to Kpler, 9.4 mb/d passed through Hormuz in the week ending 14/10, about 70% of the pre-war level.

The physical market is tighter than futures. EIA spot Brent stood at $127.43 on 09/10, while the contract traded at $103.18; we see this as a physical premium linked to scarce cargoes, without having found confirmation. Spot already exceeds the IMF's "severe" scenario, built on $110 oil and global growth cut to 2.0%, and the contract touched that level on Friday. US commercial crude stocks fell by 8.6 mb to 415.6 mb, 3% below their five-year average.

Trump's pledge not to strike Iran before 3/11 did not stop Brent from touching $110: the attacks continue and the US blockade remains in place. For currencies, the thesis of the 14/10 Daily held for oil, which rebounded after its dip towards $105, but not for the yen: lower US yields outweighed the energy bill, and USD/JPY slipped only 0.15% on the week. The oil channel would return to the fore if Brent stayed above $110 just as the US front end stops falling; the euro and the yen would then be the most exposed.

September inflation of 3.8%, the highest since September 2023 and 0.2 point above consensus, should have made the ECB more hawkish. Yet the market prices a hold on 29/10 at about 99%, and 70 of 73 economists polled by Reuters expect one. Lagarde warned at the end of September that the rise in long-term yields "will slow growth and reduce pass-through" by more than projected. With core at 2.5% and no second-round effects, the ECB can wait: 64 of 73 economists place the next hike in December.

The 14/10 Daily cited pricing of about 60% for a hike on 29/10. That figure dated from 24/09 and this week's data contradict it. The correction supports our bearish bias on the euro: the ECB offers the currency no rate support before December.

That bias was only half wrong. EUR/USD gained 0.42%, its first weekly rise after five weeks of decline, but EUR/GBP lost 0.45% and the euro also slipped against the franc: the rebound belongs to the dollar. The 10-year OAT is approaching 5% and its spread over the Bund, around 132 bp, is at its widest since 2012, after widening by more than 60 bp since August (16/10 closing level not verified).

The coming week concentrates French risk. On Tuesday 20/10 the National Assembly votes on the first part of the 2027 budget bill (PLF), which the finance committee rejected by 31 votes to 3; Moody's is due to review France's rating on Friday 23/10 (date to be confirmed; current rating and outlook not verified). The social security financing bill (PLFSS) follows on 27/10. Until these votes, our reading is that the euro will track the OAT-Bund spread more than HICP.

The three meetings follow one another from 28 to 30/10 and all are priced for a pause: about a 28% chance of a hike for the Fed according to an aggregator (CME FedWatch not consulted), about 1% for the ECB, about one in four for the BoJ based on OIS from 06/10, not updated. Yet all three signal further hikes: the risk of surprise is hawkish on all three sides.

On the Fed side, the 14/10 Beige Book adds an argument to those in the September minutes: three districts describe a "robust" rise in prices, against one in August, and four an acceleration, against one, with costs driven by energy, transport and tariffs rather than by demand. Core PCE, at 3.01% in August, still exceeds core CPI by about 44 bp. A hike six days before the midterm elections nonetheless remains difficult, and our central scenario does not include one.

The BoJ looks to us the most likely to surprise. Its rate stands at 1.25%, Ueda says the Bank intends to "continue" raising it, and board member Koeda has warned against excessive caution. USD/JPY briefly moved towards 157.5 after remarks by Deputy Governor Uchida, a measure of the pair's sensitivity to a signal from Tokyo. National CPI on 23/10, expected at 2.9% excluding fresh food after three slowdowns, will provide the final argument. At 158.02, USD/JPY prices little of this risk.

Canada lost 68,000 jobs in September, a second consecutive decline, and the participation rate fell to 64.8%, its lowest since December 1997 excluding 2020. With Brent above $100, the CAD should have benefited from its terms of trade; yet USD/CAD has gained 1.59% over a month, and its 0.46% fall this week reflects the dollar's retreat. The BoC, at 2.25%, decides on 28/10 alongside its Monetary Policy Report; Monday's CPI will show whether energy prevents it from ignoring employment.

In Australia, employment fell by 39,300 against +19,600 expected. AUD/USD touched 0.6911 on 15/10 before ending the week down 0.32%: our bearish bias of 15 and 16/10 is confirmed, narrowly. The RBA, at 4.60% according to a secondary source, now has only an 18% chance of hiking on 3/11 on Polymarket, a measure that is not OIS pricing; Q3 CPI on 28/10 will settle the matter.

China offers support that neither currency is cashing in on. September exports jumped 36.3% year on year, against 25.5% expected, and the PBoC fixed the yuan at 6.7220 on Friday, its strongest level since February 2023. Q3 GDP, expected on Monday at 4.4%, below the official range of 4.5 to 5%, will be the first test of this channel. The NZD, up 1.13% on Friday, benefited from the risk rebound more than from China; its Q3 CPI, expected at 3.7% year on year against 4.1%, is due on the evening of 21/10, Paris time (to be confirmed).