Automatic translation: Originally published in Finnish 31/05/2026, 10:38 GMT. Give feedback here.

Last week was a mixed one for the European and US stock markets. Nasdaq Helsinki declined for a change, but the US markets again aimed for new highs. Two themes continue to repeat in the markets: the effects of the Middle East conflict and the momentum of AI stocks. The price of oil decreased by 18% in May, which has been reflected in the market. In many AI stocks, nervousness seems to have turned into hubris.

I have followed with interest the news about the Fed's new Chairman, Kevin Warsh, and have written before about how Warsh wants to overhaul the Fed's operations, including the metrics it tracks. According to Warsh, inflation measures, in particular, need an update at the central bank, and he has often highlighted "trimmed mean" measures as viable alternatives to the current PCE framework. Warsh believes it is crucial to understand the underlying inflation trend rather than one-off price changes caused by geopolitical shocks or specific product categories. It is also critical to consider whether alternative measures better filter out the effects of tariffs, AI investments, and geopolitical disruptions, or if they underestimate inflationary pressures because these factors are not temporary but permanent

Source: WSJ

There are major differences between the measures. The "core PCE inflation" currently tracked by the Fed accelerated to 3.3% in April, but the Dallas Fed's trimmed mean measure showed a modest price increase of 2.3%. The latter could silence the hawks, but it also has its weaknesses. The reliability of the trimmed mean measure is overshadowed by its failure in 2021, according to the WSJ. At that time, the measure indicated that inflation was rising much slower than it actually was, leading to the Fed's incorrect interpretation of inflation as transitory. The structure of the Dallas Fed's trimmed mean measure, which removed 31% of the largest price increases but only 24% of the largest decreases, thus underestimated the inflation trend during the pandemic.

Now the same dynamics may be repeating. Tariffs are increasing the prices of goods, and AI investments are raising software and computing costs. According to researchers at the Dallas Fed, trimmed mean inflation was 0.7 percentage points lower than core PCE in April, mainly because it gave less weight to goods most affected by tariffs. Thus, bias is still observable in the measure.

Warsh's stance on new metrics is a matter of principle: the question is mainly whether the Fed can, in principle, "look through" price shocks, or if it is merely downplaying undesirable inflation figures. If tariffs and AI are permanent factors, alternative metrics provide a false sense of security, and the Fed could again be late in its actions.