Jerome Powell Favors Inflation Indicator Dropping to 2.3%, but Economists Warn It May Underestimate True Inflation

BlockBeats News, May 29th, Dallas Fed data shows that its "Trimmed Mean" inflation indicator fell to 2.3% year-on-year in April, lower than March's 2.4%. This indicator is the "trimmed average" inflation measurement method emphasized by the new Fed Chair Kevin Warsh in his confirmation hearing, which is seen by the market as a key support for his view that "inflation is improving."

However, several economists and policymakers have warned that this indicator may currently systematically underestimate the true inflationary pressures.

Dallas Fed economist Tyler Atkinson stated that due to Trump's past year of tariff policy, most commodity prices have risen simultaneously, causing the statistical distribution assumption of the trimmed mean model to fail. Under the current method, the model would exclude the top 31% of price increases and the bottom 24% of price decreases, which may artificially suppress the overall inflation reading.

Data shows that the items excluded in April include categories that saw significant increases, such as gasoline, airfare, jewelry, as well as items that experienced declines, like poultry, linen products, and haircuts. Analysts pointed out that a similar situation occurred during the post-pandemic inflation surge phase, where the trimmed mean indicator also signaled a mild inflation stance, but actual inflation was far higher than indicated.

Meanwhile, the Fed's long-preferred core PCE index continues to heat up. U.S. Commerce Department data shows that the 12-month core PCE rose by 3.3% year-on-year as of April, marking the fastest pace since 2023.

Analysts at Standard Chartered Bank stated that "it is difficult to believe that the anti-inflation reflected by the trimmed mean actually exists," and believe that this indicator, whether in historical performance or predictive ability, is weaker than the core PCE.

Harvard University economist Jason Furman also noted that the current issue is that this indicator "looks more like data picked out after the fact to support a specific view" rather than a rigorously validated consistent framework.