BEA Releases July PCE Inflation Report on Aug. 26; Forecasters Had Put Core at 3.2%–3.3%, Two Days Before Warsh's Jackson Hole Speech

The Number Comes Before the Speech

At 8:30 a.m. Eastern on Wednesday, the Bureau of Economic Analysis releases its July report on personal income and spending ^[1]. Buried inside is the number Wall Street actually cares about: core PCE, the Federal Reserve's preferred inflation gauge. Economists expect it to land between 3.2% and 3.3%, up about 0.2% from June ^[1][2][3]. Morningstar's survey puts it at 3.20%. UBS says 3.30%. Goldman Sachs says 3.24% ^[1]. Nobody quite agrees, which is itself a small tell about how murky this data has gotten.

Whatever number prints, it will sit well above the Fed's 2% target ^[2]. In June, core PCE was 3.3%, a touch below May ^[5]. That means inflation has now run hot for years, not months, and the question hanging over Wednesday's release isn't really "how high" — it's "why still."

The timing sharpens everything. Kevin Warsh became Fed chair on May 22, replacing Jerome Powell ^[9]. Two days after this release, on Friday, he gives his first Jackson Hole keynote ^[6]. Three weeks after that, the Fed makes its next rate call, on September 16 ^[6]. Warsh has also stopped including forward guidance — the Fed's usual hints about where rates are headed — in the committee's post-meeting statements, so traders have less to go on than usual ^[6]. This inflation print lands in the middle of that information vacuum, which is exactly why it matters more than a routine data day would.

A Number That Means Two Different Things

Both sides in this fight can point to the same figure — inflation running above 3% — and mean something completely different by it. That's because of a technical distinction economists call the "level versus rate" problem, and it's the real hinge of the whole debate.

Here's the mechanism. A tariff makes an imported good more expensive, once. That price increase shows up in the inflation rate for twelve months, because inflation measures year-over-year change. After a year, that same price is now the baseline, so it stops adding to the rate even though the price itself never comes back down. In theory, a single round of tariffs should look like a temporary bump, not a permanent inflation problem.

But that theory only holds if tariffs stop coming. And in 2026, they haven't. The Supreme Court struck down the Trump administration's original tariffs — built on emergency economic powers called IEEPA — on February 20, ruling out roughly 70% of the tariff structure then in place. The administration rebuilt it almost immediately under different legal authority, starting with Section 122 and adding more since, including a Section 338 tariff push against Canada in August. So the tariffs quoted in coverage now, roughly 10% to 12.5% on more than 80 countries, describe a rebuilt policy, not an unbroken one ^[9]. Each rebuild is, in effect, a new round — which is exactly the kind of event that keeps restarting the twelve-month clock instead of letting it run out.

The Fed Can't Even Agree With Itself

If tariffs were the obvious culprit, this would be a simpler story. Instead, the Fed's own researchers disagree with each other. The Dallas Fed found that tariff effects on PCE prices peaked back in the first quarter of 2026 ^[12]. The St. Louis Fed found pass-through — how much of a tariff's cost actually makes it into consumer prices — has largely leveled off since February ^[13]. The Minneapolis Fed went further, arguing tariffs simply can't explain the recent rise in goods inflation at all ^[11].

That's three regional Fed banks pointing away from tariffs as the current driver. But the Fed Board itself — the central body in Washington, not a regional bank — reached the opposite conclusion in an April staff note. It found tariffs account for the entirety of the excess inflation showing up in core consumer goods ^[14]. That finding gets far less attention in coverage than the three regional papers, even though it comes from the Fed's own house.

If tariffs aren't driving the number, something else is. Candidates include an energy shock from the closure of the Strait of Hormuz, heavy AI-related spending pushing up hardware and software prices, and portfolio management fees that rise automatically when the stock market does ^[1][9][16]. None of those respond to interest rates the way old-fashioned demand-driven inflation does, which is part of why this fight has no clean resolution.

Whose Interest Is Written Into the Argument

Warsh's case for patience isn't just about the data — it's about what a new Fed chair needs to prove. He took the job under open pressure from President Trump to cut rates, and easing now, with inflation over 3%, risks making him look like he's taking orders rather than reading numbers ^[9][15]. Removing forward guidance is part of the same logic: promising a rate path in advance makes it costly to reverse course later, so dropping the promise keeps his options open, even if it also leaves markets guessing ^[6].

The administration's case is different but not baseless. Tariff revenue and reshored manufacturing are the actual policy goals, and a one-time price bump is, in that view, a fair cost for a longer-term payoff — as long as it really is one-time ^[9]. Trump has repeatedly pushed the Fed publicly to cut ^[9]. There's also a less obvious pressure at work: higher bond yields raise what the federal government pays to service its own debt, which gives the Treasury a direct financial stake in lower rates, separate from any inflation argument. Analysts abroad have started calling this dynamic a "Bessent put," after Treasury Secretary Scott Bessent, though it remains speculation about influence rather than a demonstrated fact ^[15].

Progressive groups like Groundwork Collaborative frame it in yet another way: tariffs and a war-driven energy shock are keeping prices high while the Fed holds rates steady, meaning working households pay twice — once at checkout, once through higher borrowing costs ^[10]. Their deeper argument is distributional. A tariff functions like a consumption tax, and those tend to hit lower-income households hardest, while high rates simultaneously choke off the housing construction that could ease shelter costs.

What Markets Are Actually Watching For

Bond and currency traders mostly aren't debating causes at all — they're positioning for a jolt. Whether the Fed should raise rates in September was still unsettled heading into the release, with some estimates putting the odds of a September hike above 40% ^[3][17]. Gold traded near a 15-week high around $4,697 before slipping below $4,650 as the dollar firmed up ahead of the data ^[19]. The trade logic is simple: a soft inflation print likely pushes gold higher and weakens the dollar, while a hot one lifts Treasury yields and pressures growth stocks ^[3].

Coverage of Warsh's coming speech splits along familiar lines. A Bank of America survey found 69% of fund managers expect him to strike a neutral tone at Jackson Hole, meaning a neutral speech is already priced in — which raises the odds that any surprise, in either direction, moves markets hard ^[6]. Outlets aimed at traders, like FXStreet, lean hawkish in their framing even when the headline numbers look measured ^[2]. Groundwork Collaborative states the tariff-and-war causation as settled fact, without engaging the regional Fed research that complicates it ^[10]. Meanwhile, coverage aimed at retail investors, including The Motley Fool, wraps accurate underlying facts in more colorful language, calling this "Warsh's Trump problem" ^[9]. None of these framings are wrong about the facts; they differ in what they choose to emphasize.

One more wrinkle worth holding onto: this isn't even the first data point testing Warsh's no-guidance approach. The July producer price report came out on August 13, and it hid a 0.4% core increase beneath a flat headline number ^[16]. Wednesday's release adds to that pattern rather than starting it. And as of this writing, the actual July figure isn't public yet — 3.2% is a forecast, not a result, and whatever the BEA reports at 8:30 will be the only number that counts ^[1][2].