We’ve just published the latest update of the inflation indicators. In a nutshell, they show a steady but not overwhelming broadening in price pressures, with bond markets still more or less unconcerned, while economic experts polled by Bloomberg remain more anxious about the prospect of deflation. A rise in wage inflation and continuing pain from producer price increases, which might be exacerbated by fresh Covid-related bottlenecks, are causing the greatest immediate concern. People who argue that this inflation is transient and those who think it’s a secular trend both still have plenty of evidence. The debate remains open.
For now, I’d like to highlight one quirk of our methodology. The point of the exercise was to stick with my original selections after I’d picked the main 35 indicators. They were judgment calls, made after talking to people and giving the matter some thought, rather than quantitative decisions. That enabled us to resist temptations to “cherry-pick” to back one argument or another.
There are at least two cases where if I had made a different—and reasonable—call, we would have had a different inflation picture. Thankfully, they cancel each other out.
First, when selecting five components of the CPI inflation basket, I wanted to include a sector that might be expected to suffer inflation as the economy reopened. I chose recreation services, in blue. Recreation services inflation has picked up in the latest reading, but remains below 2% and well below its average for much of the last decade.
Had I chosen recreation, rather than recreation services, the reading would have been less benign. Inflation for the sector is only slightly higher than for recreation services—but it is running at its highest in a decade. As the heat map is driven by how each component compares with its average for the last decade, that decision made the difference between an indicator flashing a warning, and one suggesting nothing to worry about:
When it came to overall measures, I talked to a number of people about sensible gauges of core inflation. The most frequently mentioned by far was the Cleveland Fed’s “trimmed mean” which excludes the biggest outliers in either direction and takes the average of the rest. In the last couple of months, plenty of analysts have trumpeted the praises of the median (the component that sits in the middle of the distribution). I don’t recall anyone suggesting the median when I was drawing up our indicators.
There’s a reason a lot of people want to use the median now; it suggests there’s little cause for worry. The trimmed mean, on the other hand, has just risen very sharply. Here is how they’ve both moved over the last two decades, in a chart I published last week:
At the point I decided to include the trimmed mean, it suggested less of an inflationary problem than the median. Now that’s reversed.
Where does that leave us? On balance, it suggests that the anti-cherry-picking idea is working. I made my choices a few months ago. They’ve roughly evened out. So at this stage, I’m inclined to try using this method more in future. Decide what your criteria are at the outset, and do so carefully. Then stick with those criteria to help drown out the noise as time goes by. So far, I think it’s working.