The Fed's New Leader Thinks Inflation Could Use a Trim

Get personalized, AI-powered answers built on 27+ years of trusted expertise.

The Fed’s new leader thinks the statistics we use to measure inflation could use a trim.

Speaking to senators at his confirmation hearing in April, Kevin Warsh dissed the Fed’s favored inflation measure, the Personal Consumption Expenditures (PCE) price index. He said he wanted to create an inflation measure using a “billion prices” collected from the private sector. He also wants the new measure to use a “trimmed average” of prices instead of the traditional PCE.1

Changing the benchmark for the Fed’s 2% inflation target could lead to big shifts in monetary policy.

While many think of the consumer price index (CPI) when considering inflation, Federal Reserve has long favored the core PCE price index—the PCE but without food and energy prices. The idea is that removing the typically volatile prices found in food and energy, which are prone to large swings up and down in any given month, will give a clearer picture of inflation's overall trajectory and where it’s likely to go over the next few months or years.

Core PCE prices have a few problems as an inflation benchmark. For one thing, they ignore the most essential expenses in household budgets, which could create a disconnect between what statistics say about inflation and what the public is experiencing.

“If you go and you say, ‘Well, I don't care about energy and food price fluctuations,' that's like 40% of the average person’s spending,” said Stephen Cecchetti, a professor of economics at Brandeis University.

In the early 1990s, Cecchetti was a researcher at the National Bureau of Economic Research and part of a group working on a way to measure inflation that filtered out noise and showed which trends were likely to persist.

Between 1994 and 1997, Cecchetti wrote several papers with Michael Bryan, pioneering the use of measures like "median inflation" and "trimmed mean inflation" and developing indexes based on those ideas.2

Trimmed mean inflation is simple in theory: you ignore the outliers in any given month. However, experts disagree about how much of the highest and lowest prices to toss out. For example, the Dallas Fed removes the 24% of prices that fell the most and the 31% that rose the most. By contrast, the Cleveland Fed’s "trimmed mean CPI" cuts 16% total off the upper and lower tails.

There are other technical questions about whether to weight each price category by volatility or by spending, how finely to break down the price categories, and so on. All this makes trimmed means tricky to explain to the public.

“There are a lot of decisions you have to make once you start down this road,” Cecchetti said. “And while it might be useful to guide policy, the question is whether it's useful for communication, and I think that's a tougher sell.”

Emphasizing one measure over the other could affect the Fed’s interest rate policy, as the trimmed-average reading has lately pointed in a different direction than the Fed's usual inflation gauges. Trimmed average inflation has run cooler than core PCE, although historically, the opposite has been the case.

Some economists have worried that relying on trimmed averages could lead Warsh to take the risk of inflation less seriously than some of his colleagues on the Fed’s policy committee. For example, the chart above shows how the Dallas Fed's trimmed mean has run much cooler over the past year than the PCE.

“We remain skeptical about the usefulness of the measure as it tends to be slow in detecting a change in the inflation trend and has understated inflation lately,” Aichi Amemiya, senior economist at Nomura, wrote in a research note earlier this month.

Fed officials have increasingly voiced concerns that the Iran war could ignite a fresh bout of high inflation even as the pandemic-era burst of inflation lingers, keeping it above the Fed’s 2% goal for a fifth year.

However, Warsh isn’t alone on the Federal Open Market Committee in his interest in using trimmed means as an inflation benchmark. As recently as January, Fed Governor Michelle Bowman pointed to trimmed-mean inflation indexes as evidence that inflation was cooling.3

Cecchetti said the trimmed inflation measures he helped develop can help central banks filter out noise and estimate the medium-term trend for inflation, but that the Fed would be better off sticking to the CPI or PCE, not excluding food and energy, when talking about inflation to the public.

“[Trimmed mean] statistical techniques result in numbers that are going to be useful for internal decision making and technical analysis, but not necessarily that useful for public communication,” he said.

Get personalized, AI-powered answers built on 27+ years of trusted expertise.