Opinion | Inflation Is Down, Disinflation Denial Is Soaring

Sep 19, 2023 Divya Bharat

A funny thing happened to me on the internet the other day. Like many economists, I’ve been looking at various measures of underlying inflation to try to peer through the economic fog, and I thought I should post an update on one measure that I’ve emphasized in the past: consumer prices excluding food, energy, used cars and shelter. To my surprise, my dry technical post on X, the platform formerly known as Twitter, drew a huge number of mostly hostile comments:

There were also outraged comments on other social media platforms and some angry emails. The gist of most of this outburst was that I was trying to hide the problems facing Americans by using a measure that excluded much or most of what real people worry about.

Hey, if you never want to get anyone angry, don’t enter public discourse. But this was a revealing moment. Inflation has, by almost any measure, been coming down remarkably quickly and painlessly. But a lot of people aren’t having it; disinflation denial is soaring.

Before I get to that, a few facts.

First, many of those hostile commenters seem to be under the impression that economists are using special consumer price indexes that omit a lot of stuff to assess Americans’ standard of living. But nobody does that. When economists say, for example, that the purchasing power of most workers’ wages is higher than it was before the pandemic, they’re measuring prices using the whole consumer basket.

Second, many of the people weighing in seem to believe that only fancy economists’ measures show falling inflation. In fact, the plain vanilla inflation rate — measured, as in the chart above, on a six-month basis — has also plunged:

But what’s also true is that plain vanilla inflation, more generally known as “headline” inflation, is running at 3 percent, while “supercore” inflation is running at only 2 percent. Why the difference?

It’s all — actually, more than all — about housing. Here are the current levels of inflation by three measures: consumer prices excluding shelter, consumer prices that also exclude food, energy and used cars, and consumer prices in general:

It turns out that excluding food, energy and used cars actually increases the inflation measure. If I were engaged in a nefarious plot to minimize measured inflation, I’d leave them in.

Which gets to the reasons we look at inflation measures that leave some stuff out.

Why, after all, do we look at measures of inflation? One answer is that we want to assess purchasing power and the standard of living — and for that, the overall level of consumer prices is appropriate. But we also want numbers to guide economic policy, especially monetary policy. And we’ve known for a long time that headline inflation is a bad guide for policy, because it can move erratically based on clearly temporary factors. Back in 2010-11, for example, headline inflation briefly rose to almost 4 percent, leading some people on the right to accuse the Fed of “debasing the dollar” and to demand that monetary policy be tightened. But the Fed knew that this inflation was a temporary blip driven mainly by oil prices, and resisted the pressure.

The traditional way to extract the signal from the noise is to look at “core” inflation, which excludes food and energy prices. We don’t exclude these prices because we think they don’t matter, but because they fluctuate in ways that can lead to policy mistakes.

In the postpandemic period, however, traditional core inflation has fallen short as a useful measure, because it’s been influenced by some large but temporary factors. Used cars aren’t a large part of consumer spending, but their prices have fluctuated so much lately that you really want to look through their effects on inflation.

And shelter is a special problem. The Bureau of Labor Statistics measures the price of shelter by looking at rents — the actual rents that people pay if they’re renters, and an estimate of what they would be paying as renters if they own their own homes. But most renters have fairly long-term leases, so the average rent people are paying can lag far behind what new tenants pay, which is a better indicator for policy.

In the past, this wasn’t much of an issue. But there was a huge rent surge in 2021-22, probably driven by the rise in working from home. Estimates of new tenant rental rates show that this surge is well behind us, and rents may actually be falling, but the official numbers still have rents rising rapidly:

So it makes sense to drop housing from the core inflation measure, too.

Now, is what’s left after these exclusions a perfect measure of underlying inflation? Of course not. And one problem with excluding significant components of inflation is that what’s left may be driven by prices that we know are measured badly, like the cost of health insurance. But there are problems with index shopping, too; if you keep looking for better measures of underlying inflation, there’s always the temptation to pick the measures that tell you what you want to hear. Actually, one reason I made a point of posting about this measure is that I’ve used it before, so I wanted to stay consistent.

Fortunately, at this point most measures of underlying inflation are telling the same story: We’re way down, probably under 3 percent, although maybe not all the way to the Fed’s (questionable) target of 2 percent. To the extent that there was a battle among the cores, a fight over which measure of underlying inflation was best, it’s mostly over — and so, arguably, is the fight against inflation itself.

But if you say that you encounter, as I did, an outpouring of rage. Why?

A lot of it is political. Many people on the right simply assume that anyone who says that something good is happening on Joe Biden’s watch must be a liar. And some people on the left also balk at the idea that inflation is falling, because that seems to them to minimize the suffering of working families.

But not everything is political. The question of what’s happening to inflation is, or should be, a purely technical issue.