It is difficult to overestimate the time it takes to compute a consumer price index. In America, statisticians survey nearly 10,000 people every quarter, sample about 80,000 things purchased, and then monitor their prices by calling thousands of stores, restaurants, and offices. So the hard-working tinkerers might be upset to learn that the Federal Reserve believes that a better way to track inflation is simply to cut off the things with the greatest volatility. The resulting result is known as “trimmed mean inflation”. With America experiencing its toughest price pressures since 1990, this narrower yardstick is more than an academic exercise.
In August, according to data released on September 14, the consumer price index was 5.3% higher than a year ago. It was the third straight month with inflation at about this rate. In contrast, the trimmed average interest rate – using the Fed’s preferred measure, the Personal Consumption Expenditure Price Index (PCE) – stayed at only around 2%, in line with the central bank’s inflation target.
The gap between the headline horror and the far more subdued alternative brings to the heart of the debate about whether the current surge in inflation is temporary or sustained. Those who take the former argue that a small number of goods and services caused the rise in inflation, almost all of which were due to pandemic-related disruptions. For example, flight prices spiked when air traffic returned, but ticket prices plummeted in August as the Delta variant of Covid-19 dampened the desire to travel. The trimmed index is appealing because it hides such outliers. Fed chairman Jerome Powell has referred to the move as evidence that price pressures are not yet broadly based.
There are two possible objections to using the trimmed measure. The first is that the Fed is cherry-picking and emphasizing which inflation indicator looks the most flattering. Central bankers often highlighted tighter “core inflation” to better capture underlying trends. In the past, the Fed has referred to the PCE index excluding food and energy prices. This time around, however, that move is less benevolent, rising to 3.6% year-on-year in July, a three-year high. Hence the suspicion that the trimmed mean is a practical substitute.
However, this is unfair to the Fed. Central bankers followed the trimmed measure long before the pandemic. The Dallas Federal Reserve has released a version since 2005. A 2019 research note by Fed economists found that trimmed averages are less volatile than headline indices and better predict future price changes.
By focusing on the middle of the price package, the trimmed mean display also helps show how widespread price pressure is actually. (The Dallas Fed ranks all items from highest to lowest price spike, cutting off the top 31% and lowest 24% as measured by spending weights.) Trimming the mean also answers one of the standard complaints about core inflation meters that exclude food or energy prices – namely, that people spend so much money to fill up their stomach and gasoline that it makes little sense to systematically ignore these costs. Food and energy are included in trimmed indices as long as their price fluctuations are not unusually large.
The second objection is more damaging: the trimmed mean is more worrying than Mr. Powell would like. As a rough guide, Dallas Fed economists say that if headline inflation exceeds the trimmed mean by one percentage point, it will result in about 0.25 percentage points extra trimmed mean inflation in one year. On this basis, the trimmed mean rate should reach 2.5% by the end of next year. Another trimmed mean calculated by the Cleveland Fed, which is less eager in its clippings, has already risen from 2% at the start of the year to 3%.
However you trim, the conclusion seems clear enough. Inflation is not as bad as the headlines suggest. But the price pressure is spreading steadily.