The Fed's Inflation Metric Debate: Is Trimmed-Mean the Answer?

What is PCE Trimmed-Mean Inflation?

Monthly inflation data are noisy, and idiosyncratic moves by a few components can have an outsized effect. Policymakers aim to look through temporary noise and gauge the underlying trend of inflation. One of the simplest metrics that diminishes such noise is core PCE inflation, which excludes volatile energy and food prices. Most FOMC participants put a heavy weight on PCE core inflation as a gauge of the underlying trend. 

Fed Chair Kevin Warsh has criticized PCE core inflation, describing it as a “rough swag” at his confirmation hearing. Instead, he proposed alternative inflation measures, including PCE trimmed-mean inflation and private sector price measures.

PCE trimmed-mean inflation is constructed from the cross-sectional distribution of monthly price changes and is designed to remove components with large monthly swings in either direction. Currently, the metric – published by the Dallas Fed – trims 24% from the lower tail and 31% from the upper tail of the distribution. The 12-month change in PCE trimmed-mean stood at 2.35% in April, 94bp lower than 3.29% for core PCE inflation.

Problem 1: PCE Trimmed-Mean Failed to Detect the Post-Pandemic Inflation Surge

PCE trimmed-mean inflation is designed to look through idiosyncratic noise and avoid sending false positive signals about the underlying inflation trend. However, that same feature can make it less responsive when a new inflation regime is emerging, especially if the initial shock is concentrated in a narrow set of components.

This limitation was evident during the post-pandemic inflation surge. At the onset of post-pandemic inflation acceleration, annualized monthly PCE trimmed-mean inflation lagged four "true" inflation measures. On the other hand, annualized monthly core PCE inflation started accelerating at almost the same time as three true measures.

The components excluded from PCE trimmed-mean may offer a better early signal around inflation turning points. The weighted average of components trimmed from both tails began to react to the pandemic-induced inflation wave as early as Q1 2021, much earlier than PCE trimmed-mean inflation did.

Recent data show a similar reason for caution. The upper tail of the distribution has picked up, driven by higher tariffs and energy prices, and remains significantly above the pre-pandemic average (2000-2019). The lower tail also trended lower in recent months. Given the post-pandemic experience, as well as recent signs of pass-through of higher costs into consumer goods prices, the recent deceleration in PCE trimmed-mean inflation is unlikely to make policymakers comfortable.

Problem 2: A Shift in Skewness Caused Trimmed Mean to Understate Underlying Inflation

Since the last update of trimming specifications by the Dallas Fed in 2009, the distribution of the cross-sectional monthly price changes has shifted, which has likely made the current trimming proportions suboptimal.

As the distribution has become more positively skewed, PCE trimmed-mean inflation ignores the upper tail of the distribution excessively and understate the underlying inflation.

PCE Trimmed-Mean Inflation Appears to Understate Underlying Inflation by 48bp

We find that PCE trimmed-mean inflation is currently lower than the bias-adjusted trimmed-mean inflation by about 48bp on a y-o-y basis according to our estimation. This result indicates that the bias-adjusted trimmed-mean rate of inflation stands at 2.8%, narrowing the gap with core PCE inflation of 3.3% as of April 2026. Note that PCE trimmed-mean inflation and its bias-adjusted version moved in different directions in recent months.

Problem 3: Changing Goods Price Dynamics Could Complicate the Picture

A primary factor making PCE trimmed-mean inflation negatively biased is its failure to capture changes in goods price dynamics in the post-pandemic era.

Core goods price inflation, which used to be a persistent deflationary force, has consistently pushed inflation higher in the post-COVID era. This shift appears to have shifted the distribution of cross-sectional prices changes upwardly. Looking ahead, the behavior of consumer electronics prices will likely be one of key determinant factors of goods price inflation. In the 2010s, PCE core goods price inflation was almost persistently negative, led by lower prices of consumer electronics. However, many of these tech-heavy products now face upward price pressures due to global shortages of semiconductors, as well as supply-chain disruptions induced by the Iran war.

AI Investment Boom Likely to Keep Core Goods Price Inflation Positive

If the AI investment boom continues and keeps certain key components for electronics in short supply, core goods price inflation would remain positive for an extended period, which means trimmed-mean PCE inflation would continue to underestimate the underlying inflation .

Policy Implications

Dovish policymakers including former Fed Chair Powell tend to see recent inflation acceleration led by higher goods prices as temporary and maintain their sanguine inflation outlook in the medium term. 

However, persistently positive inflation of core goods prices and a resulting shift in the skewness of cross-sectional inflation distribution undermine the usefulness of certain alternative inflation measures such as PCE trimmed-mean inflation. Monetary policy driven by suboptimal inflation metrics might increase the risk of falling "behind the curve," leading to more aggressive rate hikes later.