The rate rise is aimed at a feeling, not a number

**2026-09-13 — The hike aimed at the survey.**

The August consumer price index landed exactly on forecast at the headline — 0.4% on the month and 3.4% on the year — and the annual core rate slowed to 2.4% from 2.5%. The one miss was a tenth on the monthly core. On the same day households told the University of Michigan they expect 4.6% inflation over the coming year, up from 4.0%, and sentiment fell to 47.8. Across the week from 4 September the five-year Treasury added 24 basis points to 4.79% and the S&P 500 gave up 0.80%. The Federal Reserve decides on Wednesday, and the calendar carries 4.00% against 3.75%. The print did not ask for that.

The rate rise the market expects on Wednesday is not an answer to the inflation data; it is an answer to the inflation survey. The price report gave the Federal Reserve almost nothing new — a headline precisely on forecast and an annual core rate that went down rather than up. What moved was the household's own forecast of prices, and a central bank that tightens on expectations rather than on outcomes is defending its credibility, not responding to evidence. That changes what the week is testing. The households now expecting faster price rises are the same ones whose pay is running behind prices and whose confidence just fell again, and their August spending is published on Wednesday morning, hours before the decision. Consensus expects a sharp rebound in that report. Our sharpest call in this issue is that a rebound, if it comes, will be mostly the petrol bill — and that the forecasters expecting it have not changed their view of the American consumer in six weeks, which is exactly what the proprietary layer below measures.

Start with the arithmetic of the report, because the story told about it on Thursday and Friday was not quite the report. The monthly headline of 0.4% was the fastest of the run but was the forecast; the annual headline repeated the prior month's 3.4%; the annual core slowed. Only the monthly core beat, by a tenth. Bloomberg's weekend coverage already frames energy-driven inflation as the complication in the decision, and its chief economist argues higher rates may simply be the new normal. What turned a benign report into a hike was the survey released the same day: five-year expectations moving off the level they had held since July. Central banks treat that number as the anchor, and a tenth on it is worth more to them than a tenth on the monthly core.

The household the survey describes is under real pressure. Pay rose 3.1% on the year to August against headline prices at 3.4%, the ten-year Treasury closed the week at 4.98%, and Bloomberg has rising yields pushing stocks toward correction territory. Retail sales fell in July on both the headline and the measure excluding fuel and cars. The August forecast of a 0.9% headline rebound is plausible in dollars, because fuel prices rose, and says little about volumes. Initial jobless claims at 206 thousand show the labour market is not the transmission yet; the spending line is.

Which brings the moat. The engine's consensus panel across forty US consumer-facing companies — retailers, restaurants, homebuilders, carmakers, travel groups and household staples — read at the last July snapshot and again on the settled session of 2026-09-11. On 2026-07-31 the forty names carried 1,161 buy ratings, 847 holds and 119 sells across 2,127 analyst opinions, a buy share of 54.6%. On 2026-09-11 they carried 1,153 buys, 858 holds and 122 sells across 2,133 — 54.1%. 8 buy ratings changed hands in six weeks. Over the same six weeks Michigan sentiment went from 55.2 to 47.8 and its expectations component printed 45.8. The engine's insider-cluster series moved where the panel did not: the week of 27 July logged 154 distinct buying clusters against 61 selling, about 2.5 to one; the week to 2026-09-11 logged 180 against 150, about 1.2 to one.

An analyst rating is a published model of a company's next few quarters, and every model on a consumer-facing company is built on a consumer. When the household's own reading of its prospects falls this far and the ratings on the businesses that sell to it barely move, one of the two is stale — and it is not the one refreshed every month by the people doing the spending. The panel is the company-level twin of Wednesday's forecast for retail sales: both still describe the shopper as it looked in July.

The insider series is the discriminating observation, because the people running these businesses see the till before the analyst does. Had the slide in confidence been a mood about petrol prices that never reached a checkout, executives would have kept buying their own shares at the summer pace. They did not: buying clusters are down by roughly a third from their August peak and selling clusters have more than doubled since late July. That series spans every sector rather than consumer names alone, and the consumer clusters inside it are mixed, so it corroborates rather than proves.

The read that follows is about where the adjustment lands. If spending holds, the analysts were right and the survey was noise. If it does not, the correction arrives as a round of estimate cuts across a panel that has not yet begun making them — after the rate rise is already in place.

Step back from Wednesday and the structure of the week is that the United States is setting the world's price of money against its own household's expectations, while much of the rest of the world spent the weekend organising around that price. In New Delhi the BRICS members agreed to widen trade and payments in local currencies, Xi Jinping made his first visit to India in seven years to reset ties with Narendra Modi, and Iran's president met the Abu Dhabi crown prince on the summit's sidelines. Canada, meanwhile, is canvassing a trillion dollars from investors looking for a haven from Washington. None of that reprices anything on Monday. What it does is lower the cost, for everyone else, of not following the Federal Reserve: Brazil's central bank is expected to cut to 13.75% on the very day Washington is expected to hike. A currency that raises its own price while its trading partners build alternatives is running a credibility test in two directions at once — at home on expectations, abroad on usage.

This read is wrong if Wednesday's spending report beats on the control group — sales excluding petrol stations and cars — rather than on the fuel line. A household that is spending in real terms while telling surveyors it is miserable is a household a rate rise can lean on safely, and this issue will have mistaken a mood for a balance sheet.

**Methodology.** Tape figures are quoted from the settled session of 2026-09-11 against a fixed base of 2026-09-04; bitcoin and ether are also quoted at their completed Saturday daily bar, and no window in this note ends on the publication date. The August consumer price figures are now in our calendar and are quoted as ours; the 12 September edition cited them to outlets and described the consumer core as hardening, which holds on the month and not on the year. That edition also carried the German ZEW consensus at 21 and described an expected thirteen-point fall; the calendar has since been revised to 37 against 34.2, and that framing is withdrawn. The consumer panel is a fixed list of forty US consumer-facing companies read from the analyst consensus snapshots on 2026-07-31 and 2026-09-11. Insider clusters count distinct cluster keys per week across all sectors. Port and tanker baselines exclude feed-outage zero days. Positioning is taken from the highest-open-interest contract market per commodity, report dated 2026-09-08.