Extending the Range of Robust PCE Inflation Measures

TL;DR

This paper evaluates various robust PCE inflation measures from 1970 to 2024, finding that median and trimmed mean measures outperform core inflation in forecasting, and proposing the use of a range of measures to better track trend inflation.

Contribution

It systematically compares multiple robust inflation measures, revealing the advantages of using a range of trims over a single measure for trend inflation estimation.

Findings

Median and trimmed mean inflation outperform core inflation in forecasts.

No single trim is consistently best among alternative measures.

Using a range of inflation estimates improves trend inflation tracking.

Abstract

Robust inflation measures gauge inflation behavior by excluding volatile expenditure categories from headline inflation. We evaluate the forecasting performance of a wide set of such measures between 1970 and 2024, including core, median, and trimmed mean personal-consumption-expenditure (PCE) inflation. Core inflation performs significantly worse than official median and trimmed mean inflation. Among a set of alternative trimmed mean measures, there is no single best trim based on forecasting performance: A wide set of trims generates statistically indistinguishable average errors. Nonetheless, different trims imply different predictions for trend inflation in any given month, within a range of 0.5 to 1 percentage points. In tracking trend inflation, this range and its midpoint outperform all trimmed mean inflation measures, suggesting the use of the range of inflation implied by the set of near-optimal trims as a valuable complement to any single inflation measure.