Futures markets now put the odds of a quarter-point Federal Reserve hike this month at 66%, and almost every explanation you will read today points at the same thing: Brent crude sitting near $95 a barrel, tankers under fire near the Strait of Hormuz, and an energy shock feeding straight into American prices. That story is intuitive, it is dramatic, and it is not what Fed Chair Kevin Warsh actually cited when he moved the market.
In his Jackson Hole keynote on August 28, Warsh recommitted to the 2% target measured in personal consumption expenditures. That single choice is doing more work than the oil price. Core PCE ran at 3.3% in July. Core CPI, the other headline measure of underlying inflation, ran at 2.5%. Those two gauges are supposed to tell roughly the same story about the same economy, and right now they are 80 basis points apart. The case for hiking in September rests almost entirely on which of them you decide to believe.
Two Numbers, One Economy
The divergence is not a rounding artifact. The Bureau of Economic Analysis reported that core PCE held at 3.3% annually in July, unchanged from June, with the retreat toward 2% stalled well above target. The Bureau of Labor Statistics put core CPI at 2.5% over the same period, with headline CPI at 3.4% after a 0.1% monthly rise.
Historically the two run close, with PCE typically printing 30 to 40 basis points below CPI. That relationship has inverted. Analysts at PIMCO, writing on why the two US inflation measures now tell different stories, have flagged that the year-over-year gap has swung to roughly positive 60 basis points, one of the largest reversals since 1985. In July it was wider still.
So the Fed is not choosing between “inflation is 3.3%” and “inflation is 2.5%” as a matter of precision. It is choosing between two defensible statistical constructions that point at opposite policy conclusions.
Why the Gap Exists
The mechanics are unglamorous and they matter enormously.
Shelter is the biggest single driver, as Belgian bank KBC has laid out in its breakdown of why the two US series diverge. CPI assigns roughly 36% of its basket to housing costs, including the imputed rent economists calculate for owner-occupied homes. PCE assigns closer to 15%. With shelter inflation still sticky around 5.4%, that weighting difference alone pushes CPI’s headline up while doing comparatively little to PCE.
Running the other way, PCE carries heavier weights in medical care services, financial services and software, categories where prices have climbed sharply through 2026. PCE is also broader by construction: CPI captures what households pay out of pocket, while PCE captures all consumption regardless of who writes the check, which is why employer-funded and government-funded healthcare shows up far more forcefully in the Fed’s preferred number.
Strip that down and you get an uncomfortable summary. Core PCE is elevated substantially because of services inflation concentrated in healthcare, finance and software. It is not elevated because of crude oil.
The Oil Shock Is in the Number the Fed Ignores
This is where the popular explanation falls apart. Energy is not in core anything. It sits in headline inflation, which is precisely the measure central bankers spend their careers explaining they look through.
And the energy data cuts against the panic narrative. Energy costs actually fell 1.5% month over month in July, led by a 2.9% drop at the pump. The crude spike that has driven this week’s coverage, with Brent trading near $94.86 and US strikes around Hormuz answering Iranian mining attempts, is a September story. It is not in the July prints that Warsh was describing at Jackson Hole.
An oil shock raises the price of a barrel. It does not raise the price of a hospital stay, a brokerage fee or an enterprise software seat, and those are what core PCE is actually measuring.
The transmission channel everyone is watching is real but slower and less certain than the headlines imply. Meanwhile the US 10-year Treasury yield has pushed to 4.80%, its highest since January 2025, repricing the cost of capital across the entire economy well before the FOMC votes on anything.
Warsh Picked the Gauge That Licensed the Move
Here is where BusinessTech.News will say something the wire copy will not.
Warsh did not stumble into the PCE framing. He is a deliberate communicator operating under direct political pressure from a President demanding lower rates, and he chose, in a set-piece speech at Jackson Hole, to anchor on the one gauge that reads 3.3% rather than the one that reads 2.5%. Bond traders understood immediately: as CNBC’s analyst roundup from the symposium recorded, hike expectations jumped from roughly a third to 55% the same morning, and to 66% by the end of the month.
That is a defensible choice. PCE is the Fed’s stated target, it has been since 2000, and switching gauges when the preferred one turns inconvenient would be far worse. But defensible is not the same as explained. The Fed has not told the public why a measure inflated by shelter-weight mechanics and healthcare imputation should override a measure sitting three-tenths from target, and it has not acknowledged that the gap is near a forty-year extreme. Warsh spent his speech insisting the summer’s better readings did not convince him that underlying trends had improved. He did not spend it explaining why the two official readings of those underlying trends disagree by the widest margin most working traders have seen.
If the Fed hikes on 3.3% while 2.5% goes unmentioned, it owes the country the reasoning. Raising the cost of capital into a supply shock is a decision with victims, and the borrowers already repricing are not the ones mining the Strait of Hormuz.
What to Watch
The Beige Book lands today alongside the August ADP employment print, and Friday brings the jobs report that has already scrambled these odds once. Back in March, when hike odds first crossed 50%, the framing was identical: oil-driven inflation forcing a hawkish turn. Six months later the crude price is higher, core CPI is lower, and the gauge gap is wider.
Watch whether any Fed governor is willing to say the quiet part before the September meeting. If the committee hikes without addressing the divergence, it will have made the single most consequential monetary decision of the year on a measurement choice it declined to defend in public.