Against a sectoral approach to analyzing inflation
On Wednesday morning, the Bureau of Labor Statistics will report the much-awaited April inflation numbers. After March’s extreme 1.2% (month-over-month) print, it seems highly likely that the CPI will be much lower, perhaps around 0.3%. Five-year breakeven inflation rates (which are derived from the spread between inflation-protected Treasury rates and unprotected rates) have now fallen to their late February levels, and indicate a medium-term annual inflation rate of about 2.3%. While it’s not quite hitting the Federal Reserve’s flexible average inflation target that they announced in 2020, these numbers are fully consistent with low, stable and well-anchored inflationary expectations. All in all, this is quite good.
But I think this (likely) rapid fall in inflation might lead to some of the same poor epistemic practices that led many pundits to under-rate how high inflation would rise in the winter. Specifically, I would like to caution against the practice of citing specific industries to explain why inflation is “really” higher or lower than the headline number.
Recall autumn 2021, when inflation started rising for real. Many pundits would try to rationalize the high prints by explaining that if you remove one or few outlier sectors (usually used cars), inflation was actually at a manageable level. Of course, used cars turned out to be a canary in a coal mine and hardly an outlier at all. Why is this? Because inflation is not just a series of specific price increases, but rather an increase in the overall price level. If the Fed appropriately manages nominal spending (which they have near complete control over), then adverse supply shocks in specific sectors like used cars will mostly cause relative price changes, but not a major change in the overall price level. Depending on how the Fed sets its nominal spending, when a drought hits Kansas and severely reduces the supply of wheat, the price of wheat will rise but other prices and wages will fall as people have less money to spend on other goods. The overall price level will change at the exact rate that the Fed determines based on their control of nominal spending.
Consider a toy example. One morning, every American wakes up with an extra $10,000 in their pockets courtesy of a magic fairy. Many will rush out to spend it. Presumably, some sectors will see a lot larger increase in demand than others. Inelastic goods like fuel and food would probably see a modest increase, while more discretionary purchases like vacations would probably see a major uptick. As a result, the price increases would be highly unequal. But would it be accurate to say “well, 70% of the inflation is caused by a few outlier sectors and therefore real inflation is actually much lower”. Not at all–focusing on individual sectors entirely obscures the fact that the cause of the overall price increase is the change in nominal spending caused by the magic fairy, not any industry-specific factors.
At the risk of belaboring the point, I’ll throw in one last example. Suppose a Disney craze strikes America and everyone wants to go to Disney World, buy tickets to Disney movies and buy Disney toys. Will this cause inflation? No. Certainly the price of Disney products will increase substantially. But because people spending more on Disney products means they have less money to spend elsewhere–so long as the Federal Reserve keeps overall nominal spending trajectory constant–then other prices will decline and the inflation level will not change. In short, inflation is simply a function of nominal spending.
This is why it was mildly frustrating to see smart individuals last autumn tell us that inflation wasn’t so bad last autumn because it was mostly used cars. Those among them who are intellectually honest (and, to be fair, most of them are) would then have to acknowledge that the inflation prints now will be understating inflation because much of it will be driven by large outlier drops in specific sectors like used cars.
There is one very large exception to all of this: food and energy. These items are generally excluded from calculations of “core” CPI because they are highly volatile and are not predictive of future months’ inflation print. That doesn’t mean that food and energy prices don’t matter, but if large sticker prices are caused by a jump in the price of oil (like was true in March), that probably means that the inflation number will be lower next month (as will all but certainly be true in April). So people excluding those two sectors are totally fair to do so. If anything, forecasts suggest that headline CPI (which includes food and energy) will be below core CPI in April precisely because of the steep drop in the price of oil.
There is a healthy way to cite individual industry price numbers responsibly, but it’s rather mundane. Sometimes industries publish price numbers before the BLS and analysts will cite those as a weak crystal ball. This is entirely legitimate and good. But it’s important to remember that inflation is not caused by a series of industry-specific factors, but rather by changes in the overall price level. Those who rattle off the idiosyncrasies of each industry to explain inflation merely give the illusion of expertise, but only help muddle the debate. Focus on nominal spending and market-based forward-looking indicators and everything will become clearer.