There’s a new inflation measure in town. Trimmed-mean inflation is in the spotlight after US Federal Reserve Chair Kevin Warsh said at his confirmation hearing that he preferred to track inflation using trimmed-mean measures.
Calculating a trimmed mean is fairly straightforward. Take the price changes for each item in the inflation basket over a given period, arrange them in ascending order, trim away the outliers at both ends, and calculate inflation using the remaining data. The central portion left after the most extreme values have been excluded represents the underlying trend in inflation that is less affected by erratic, one-off price changes.
As a concept, it is elegant. As a measure, it is familiar: the Dallas Fed and the Cleveland Fed both calculate and publish trimmed mean series based on different underlying inflation indices.
As a policy tool, it brings a different perspective to inflation analysis.
The key question is: how much to trim? The Dallas Fed lops off items whose price changes are below the 24th percentile or above the 69th percentile of the distribution. The Cleveland Fed does a symmetric trim, cutting 8% from both the top and the bottom—so 16% of price changes are excluded. The Reserve Bank of Australia trims 15% from each end. Other inflation-targeting central banks, the European Central Bank, the Bank of Japan, the Bank of Canada and the Bank of England, also monitor exclusion-based measures to gauge inflation trends.
The Reserve Bank of India neither publishes nor refers to trimmed mean data in official statements. But since the Fed chair has highlighted it, the measure is likely to attract interest soon, at least among economists and market analysts.
India’s Case
It is therefore useful to construct a simple trimmed-mean inflation series for India using the new CPI series and two different trimming thresholds—8% and 15%, applied symmetrically to the top and bottom. The series is necessarily short: item-wise data under the 2024-base CPI series are not available before January 2025, so year-on-year inflation can be calculated only from January 2026, leaving just six months of trimmed mean data.
Between January and June 2026, food, household appliances, motorcycles and scooters dominated the set of items trimmed from the lower end, as they recorded the smallest price increases over the period. Items excluded from the top included cooking oils, gold, silver and other precious metals. Petrol, kerosene, diesel and airfares were added as the impact of the war in West Asia fed through to prices. Trimmed mean inflation was lower than headline and core inflation in the first six months of 2026. However, the gap with core inflation narrowed in June due to rising food prices. When food prices rise, more food items are excluded from the trimmed mean, bringing core and trimmed mean inflation closer together.
The RBI’s flexible inflation-targeting mandate requires it to maintain headline CPI inflation in the range of 2–6%. Given this clear mandate, is there any advantage in generating yet another inflation number?
Yes, because policymakers are grappling with unprecedented uncertainty and need every possible tool to analyse and predict inflation. Climate change has led to heat waves and floods that could seriously affect global food prices; indeed, India’s own food output is closely linked to the effects of El Niño. At the same time, the continuing war in West Asia threatens fuel-price stability, while the AI boom is expected to push up the prices of chips and computing infrastructure. With so many scenarios in play, it is critical to separate noisy data from trend signals. This is where trimmed mean inflation does a commendable job. In fact, because it excludes the items with the most volatile prices rather than a predetermined group, it is a good proxy for underlying core inflation.
Policy Trade-offs
Policymakers are already providing more detailed inflation information to the public. In its monthly CPI press release, the Ministry of Statistics and Programme Implementation has started publishing a list of the top five items with the highest and lowest inflation rates for that month. Highlighting the most volatile items in the inflation basket is closely aligned with the idea of a trimmed mean.
In recent years, RBI officials have often referred to changes in core inflation and core ex-gold to explain their inflation outlook. In other words, inflation data are already being sliced and diced in many ways to facilitate policymaking.
For all the value of expanding the inflation line-up, there are downsides too. The risk is that trimmed-mean inflation—which is typically lower than the headline rate—may become a convenient excuse for avoiding rate hikes.
With so many inflation numbers available—wholesale-price, purchasing-price and services inflation—it becomes easier to shop around for the measure that justifies a policy decision. This is less likely in India, where CPI is specified as the inflation target, but such fears are being voiced in the US. The deeper concern is that trimmed-mean inflation is even more divorced from “felt” inflation than headline and core inflation are.
The average household cannot interpret an inflation number generated by excluding a different subset of outliers each month; it has little connection to lived inflation. People cannot simply trim the prices they pay, nor can they stop consuming all high-inflation items. Measures such as trimmed mean inflation are therefore best left to experts and researchers.
.png)
