Core PCE stalls at 3.35%, and the September cut nobody is pricing

Start with the weight. Over 60% of consumer spending goes to core services, a bucket that includes rent, healthcare, insurance, transportation services and subscriptions ^[8], and that bucket ran 3.7% above a year ago in July ^[7]. Multiply the share by the rate and core services alone deliver roughly 2.2 percentage points of consumer price growth ^[22], more than the Fed's entire target. Printing 2% on core PCE therefore requires almost everything else to be flat or falling. It is not. Durable goods rose 0.37% in the month, an annualized 4.6%, and 3.4% over the year ^[10].

Wolf Richter's explanation for the services stickiness is a pricing-power argument rather than a demand argument: many core services either lack the competition consumers can leverage or cannot be shopped around at all, healthcare being the obvious case, so firms that believe they can raise prices do ^[9]. That is not a category that responds quickly to a policy rate.

The newer line item is the one worth watching. The PCE index for information processing equipment, meaning computers, tablets, accessories and software, rose 1.4% in July, 15.5% over the year, and 22% across the eight months since the break began in December ^[11]. Compounded, that eight-month run is about a 34.8% annual pace ^[24]. Richter attributes it to the AI boom reaching consumers directly ^[11], and the mechanism looks like the one already visible in jewelry and watches, up 14.8% year over year as multi-year gold prices get passed through ^[12]. Upstream cost, retail price, core index.

The result is a Fed pinned by its own yardstick. Core PCE has been above 2% since March 2021, bottomed at 2.6% in April 2025, and has moved away ever since ^[6], which puts July 0.75 points above the low ^[21] and 1.35 points above target ^[20]. Federal funds futures tracked by CME Group had just under 60% of traders pricing no change in September and about 40% pricing a hike ^[15]. Those two add to essentially the whole distribution, leaving no meaningful probability on a cut ^[23]. The easing debate is not being lost. It is not being held.

What American Banker reported as the market's consolation was an advance retail sales reading down 0.6% in July ^[18], welcomed as a sign the Fed may be less inclined to raise rates next month ^[19]. Note what that argument rests on: weaker spending, not better prices. Governor Lisa Cook, meanwhile, said that if she does not see signs of continued disinflation soon, she is prepared to act ^[16].

One small discrepancy is worth flagging for anyone modelling off headlines. American Banker put core at 3.3%, unchanged from June ^[3]; Wolf Street put it at 3.35%, also unchanged ^[2]. The half-tick is rounding ^[25], but it is the difference between a number that reads as drifting toward 3% and one that reads as parked. The monthly data says parked: 0.25% in July, a 3.0% annualized pace ^[1]. And the households doing the paying do not get the Fed's privilege of looking through energy, still up 15.3% over the year with gasoline up 25% ^[13].