THINK Ahead：为什么凯文·沃什对通胀的看法是错误的

The Fed Chair made it clear at Jackson Hole that underlying inflation is not improving. But this is simply not what the data is telling us, writes James Smith. A rate hike may be coming, but the case is far from compelling. Read on for our guide to the week ahead

There can now be little doubt about Kevin Warsh's views on US inflation.

Every syllable of his Jackson Hole speech last week was hawkish. The core message was that he’s a big-picture guy. And the big picture on inflation, he argued, simply isn’t getting any better. But is that really true?

Take this chart, the centrepiece of Warsh’s speech:

This shows the distribution of inflation rates across roughly 200 categories within the Fed's preferred inflation gauge, the core PCE deflator. More than half of those categories are currently experiencing inflation above 3%. A quarter are running above 5%.

That sounds alarming. Yet I have some problems with all of this.

My main bugbear is that it purposefully avoids weights. Doctors' services, which account for roughly 4% of the inflation basket, are treated with exactly the same importance as intercity buses, which account for around 0.006%. A lot simply depends on how granular the statisticians have decided to report different pricing categories.

What’s more, when you break it all down, you find that goods prices account for virtually all of the worrying shift in the distribution of price increases. Many of these goods have small weights in the overall basket. And more importantly, it suggests tariffs are the main culprit behind Warsh’s scary chart.

But the impact of tariffs is changing rapidly. Following the Supreme Court's ruling against large parts of President Trump's emergency tariff programme earlier this year, tariff refunds now exceed tariff revenues collected by the US Treasury. In July alone, refunds amounted to roughly 10% of the value of total US imports.

Not all of these savings will be passed on to consumers. But the direction of travel is clear. Tariffs are unlikely to be a source of inflationary pressure for much longer. Warsh’s chart should start trending lower.

None of this means inflation concerns should be dismissed altogether. The share of service sector categories running above 3% inflation remains well above pre-pandemic norms and has shown remarkably little progress.

But here too, there are issues. It’s actually quite difficult to measure services inflation – and around a fifth of the services prices aren’t measured at all. They’re “imputed” – the statisticians have to find a proxy. Financial services account for most of this.

I bring this up because influential Fed Governor Chris Waller said this week that he ignores these imputed prices completely. They don’t tell us much about the true inflation rate facing households. Strip them out and so-called market-based core inflation currently runs roughly 0.3 percentage points below the headline core PCE figure of 3.3%. In July alone, roughly half of the month-on-month increase in prices was down to these imputed categories. That’s not exactly trivial.

So if Warsh’s measure of inflation is flawed, what should we look at?

One option is to look at “trimmed mean” inflation. This strips out the categories with the highest and lowest inflation rates at any given time, cutting out some of the noise in the charts I shared earlier. We know Warsh himself has warmed to this indicator recently.

And it’s striking. Trimmed mean inflation has fallen in 2026 and now sits pretty close to the Fed’s 2% target, in sharp contrast to what most other measures are telling us. Again, I suspect some of this is down to tariffs.

Now without boring you even further into submission, this isn’t a problem-free measure of inflation either. So maybe we shouldn’t try to get too clever – and simply look at what the headline metrics are telling us.

Here, there is room for optimism. The six-month annualised rate of core PCE inflation, a decent sense of the underlying trend, has started to turn lower after two cooler prints this summer. On a three-month basis, the news is even better. And it's even better still if you look at the alternative CPI measure of inflation.

This was the point Waller made this week; he argued that barring another hot reading, he’d be voting for a ‘hold’ this month.

What’s more, some of the pick-up in that six-month metric earlier this year might simply be down to dodgy seasonal adjustment.

Ever since 2023, core PCE inflation has been higher in the first six months of the year than the second, and especially across January and February. This year has followed a broadly similar pattern – particularly in the first quarter.

This shouldn’t be happening, but it is. And it points to cooler inflation readings in the second half of the year which should drag down that six-month average even further.

The simple message here is that the inflation backdrop is not as bad as Warsh made out last week. And Waller’s view feels more in tune with the latest data.

This is without talking about the range of other factors my colleague, James Knightley, points to as reasons for relative optimism on inflation. Think about declining rental growth. And a muted wage backdrop.

Let us not be naïve. It is unlikely Warsh would come out with such a hawkish mantra and not vote to hike interest rates. That is now our view, unless something really crazy happens in the data. Today’s bumper payroll data suggests that’s unlikely.

But the case for raising rates simply because of recent inflation data is far from clear-cut. It’s the same conclusion I reached for the eurozone last week. And it’s why James Knightley thinks a Fed rate hike might end up as a one-and-done.

If you're still hungry for more, tune into our webinar next Wednesday where James Knightley and I, among others, will be talking about all the big central bank meetings this month. Sign up today!

• The outcome of the September FOMC meeting remains in the balance. In the wake of Kevin Warsh’s Jackson Hole Symposium address, we changed our view to a rate hike being more likely than not given the tone of comments – the focus on inflation in an environment of full employment and where financial conditions are not regarded as tight. With few on the Fed openly hostile to a hike and with Treasury Secretary Scott Bessent and even President Trump seemingly on board, we characterised it as a risk management move, similar to that enacted by Alan Greenspan’s Fed 30 years ago in 1996.
• Aug Inflation (Fri): Critical to the outcome will be the August inflation print, published on Friday 11 September. Headline inflation is going to be hot given the moves in energy costs. We look for a 0.4%MoM outcome, but the core measure should be a more modest 0.2%MoM. That elevated headline number, coupled with further rises in energy prices due to ongoing issues surrounding the Middle East, runs the risk of some further pass through into core inflation via freight charges and airline fares. We expect this to be enough to nudge the Fed into a precautionary hike.
• ECB rate decision (Thu): The stage looks increasingly set for another rate hike at next week's European Central Bank meeting. Not only because some ECB members already advocated for a rate hike at the July meeting, but also because, since then, the eurozone economy has shown an almost unexpected resilience to the war in the Middle East. At the same time, headline inflation has continued to edge higher and looks set to stay above 3% year-on-year for the remainder of the year, even if other inflation measures like core and services currently provide no reason to panic. Whether the ECB will really go beyond a September rate hike is a completely different story. Read our full preview