August PCE Inflation: 3.4% Overall and 3.0% Core, Below Forecasts, After Bureau of Economic Analysis Method Changes Cut July's Readings

The Number That Didn't Move

On September 30, 2026, the Commerce Department released its Personal Income and Outlays report for August^[1]. The headline measure, the personal consumption expenditures (PCE) price index, rose 3.4% from a year earlier^[2][3]. That's the Federal Reserve's preferred inflation gauge. The "core" version, which strips out volatile food and energy prices, rose 3.0%^[2][3].

Both numbers beat the forecast. Economists polled by LSEG had expected 3.7% headline and 3.3% core^[2]. That's a real gap on paper. But here's the catch: August's reading was flat against July. Nothing cooled between the two months. Instead, July's own number had just been revised down, from 3.7% to 3.4% overall and from 3.3% to 3.0% core^[2][7].

So the "surprise" didn't come from prices falling in August. It came from the ruler changing length the same day the measurement was taken. That single fact splits the entire coverage of this report down the middle.

Why the Ruler Changed

The Bureau of Economic Analysis, which compiles the PCE index, launched its 2026 annual update of the national accounts on the same day it released the August data^[1]. As part of that update, it changed how it measures prices in three categories: legal services, computer software, and portfolio-management fees^[5][6]. It applied the new method all the way back to 2021^[5][6].

The BEA didn't do this quietly or without reason. It said the old data source for legal-services prices had shown "erratic changes that cannot be corroborated"^[5]. In plain terms, the numbers feeding that part of the index had started jumping around in ways nobody could verify or explain. The agency replaced them with producer price data and new composite indexes for software and money-management fees^[5].

Revising old data when a measurement method improves is standard statistical practice. It keeps the series internally consistent, so a reader comparing this August to last August isn't comparing numbers built two different ways^[1][6]. The problem is one of timing, not technique: the fix landed in the middle of a period when every inflation report gets read for signs of what the Fed will do next^[1][5][6].

Two Ways to Read the Same Report

This is where the story splits. One camp says the revision simply corrected bad data, and the corrected numbers show real cooling underneath. On the revised figures, core prices over just the last three months have been rising at a 2% annual pace, according to the macro research group Employ America^[9]. That's a pace right at the Fed's target. If that holds, it suggests the underlying trend is better than the headline year-over-year number implies, and that raising rates further than needed could slow the economy for no reason^[8][9].

The other camp, including the financial blog ZeroHedge and the libertarian Wichita Liberty blog, says nothing about actual prices changed at all. ZeroHedge wrote that the lighter-than-expected result was "likely not due to lower prices but due to spreadsheet changes and rebenchmarking"^[6]. In their reading, inflation held flat at 3.4% against the revised July number, and the monthly increase in prices actually picked up as energy costs climbed^[6][11].

Both sides are looking at the same release. What divides them is how much they trust the new method, and how much weight they put on the energy price increases still showing up in the monthly data^{[4][6][9][11]}. The Conference Board, a business-funded research group, sided with the skeptics' conclusion, if not their framing, titling its note on the report "August PCE Portends Further Hikes Remain Likely"^[10].

The War in the Gas Pump

One thing both camps agree on: energy costs rose again in August, and that's tied to the ongoing Iran war^[4][15]. That same pressure pushed the separate consumer price index (CPI), a different but related inflation measure, up to 3.4% for August as well^[4][15]. CNN led its coverage with this angle, framing U.S. inflation as "stubborn" and pointing to the war as the driver, rather than leading with the revision^[4].

That's a real and defensible read of the data. Energy prices move every household's budget whether or not an economist is debating methodology behind the scenes. But it also means two true things are happening in the same report, pulling in opposite directions. Headline measurement is lower because of a methods change. Monthly cost pressure from energy is higher because of a war. A single news hook can only lead with one of them.

Underneath both arguments sits a harder number: households are spending more than they earn, and have been for months. Inflation-adjusted spending rose 0.6% in August, the strongest monthly gain in over a year. Income rose only 0.2%^[1]. To cover that gap, people are saving less. The personal saving rate fell to 4.1%, the lowest level since November 2022^[1][6]. That's the cushion families would lean on if prices jumped again or a job was lost, and it's thinner than it's been in almost four years.

What the Fed Does With This

Markets reacted to the report by trimming the odds of an October rate hike. The CME FedWatch tool put those odds at 37% after the release^[3]. The Fed, which already raised rates in September, now has to decide how much weight to give a report that's genuinely ambiguous about whether inflation is actually easing^[3][8][16].

Fed hawks, meaning officials and analysts who favor keeping rates higher to fight inflation, note that even the revised 3.0% core figure sits well above the Fed's 2% target^[3]. A month of better-looking data, built on a changed yardstick, isn't proof that the underlying problem is solved^[10]. Cutting or holding rates too early, in their view, risks letting high inflation become something households and businesses simply expect and build into their own price-setting^[8][10].

The other side points to that same 2% three-month core pace as the more honest signal, arguing it reflects what inflation looks like with the bad legal-services and software data stripped out^[9]. Markets split the difference for now: December is still seen as the more likely date for another hike than October^[3][16]. Which interpretation holds up will come down to whether next month's report, built entirely on the new method, confirms the cooling or confirms the energy-driven heat.