It’s going to be a year of bottlenecks and blips.
This article is a wonky edition of Paul Krugman’s free newsletter. You can sign up here to receive it.
If Thursday’s retail sales report is anything to go by — and it is — we’re about to see a really big boom. Between stimulus checks and vaccinations, we’re very likely headed for a year of growth faster than anything since 1984. Happy times are here again!
But what about inflation? The debate we’ve been having over whether the American Rescue Plan is excessive is kind of surreal for those who remember the macro debates after 2008; this time, economists on the other side are neither knaves nor fools. There are indeed reasons to be worried about inflationary overheating. In fact, even those of us who think it will be OK expect to see above-normal inflation this year. We just think it will be a blip.
What do I mean by that? In 2011-12 it was fairly easy to debunk inflation worries by pointing to “core” inflation, which excluded volatile food and energy prices. Obviously we’ll be looking at the same measure this time. But there are reasons to think we’ll see a transitory surge in core inflation too, which doesn’t represent a deeper problem.
And it seems to me that we should make that argument now, so as not to be accused of making excuses after the fact. This is a good time to identify which aspects of inflation might worry us, and which shouldn’t.
So let’s talk about why we needed a concept like core inflation to begin with, and why we might need an extended concept — supercore? — this time.
Inflation: It’s all about the inertia
Most official U.S. economic data extend back only to 1947 (labor markets, income distribution) or 1929 (G.D.P. and all that). Consumer price data, however, goes all the way back to 1913. So we can take a very long view of inflation, which looks like this:
Spikes in inflation aren’t a new thing. There was huge inflation during World War I; there were bursts of inflation during World War II, after the war when price controls were lifted, and again during the Korean War. However, all of these inflation surges were brief. It wasn’t until the 1970s that we got an extended period of high inflation.
And unlike previous bouts of inflation, the 70s inflation was sticky: It didn’t go away as soon as a wartime boom was over. Instead, inflation became embedded in the economy, so that bringing it down required putting the economy through a wringer. Paul Volcker slammed on the monetary brakes in 1979, not taking his foot off until 1982, and the economy went through years of very high unemployment:
But wait: I just used evocative language without really explaining what it means. What are we talking about when we talk about “embedded” inflation?
The somewhat paradoxical answer is that embedded inflation, which is the kind of inflation we really need to worry about, is inflation in prices that don’t change very often.
Inflation persistence: The staggering truth
Some goods, notably things like oil and wheat, have constantly changing prices. But many don’t. There’s a large economic literature on how often prices change (here’s a summary); it’s important to distinguish between changes in base prices and occasional sales. When you do that, you find that many firms are reluctant to change prices too often; the median consumer price changes only around once every 7-11 months. And wages and salaries are normally set for a year.
Why is this true? Why does the frequency of price changes vary so much across goods? Those are deep, hard questions, which I have no intention of trying to answer.
Instead, let’s focus on the consequences of intermittent price adjustment, which takes place in a staggered way — that is, all prices don’t change at the same time.
Imagine an individual seller which changes prices infrequently, say once a year, but tries to keep its average price over time in line with costs and the prices charged by competitors. And imagine that this seller has been operating for a while in an environment in which the overall level of prices is rising at a moderately fast clip, say 10 percent a year. Then this seller’s price will look like this over time:
That is, each time it resets the price it will mark it up both to make up for past inflation and to get ahead of expected future inflation. If it changes prices once a year, it will raise the price 10 percent on each reset.
And if there are lots of price-setters acting this way, it means that overall prices will rise at 10 percent a year, even if there’s no new inflationary pressure — that is, even if supply and demand are balanced and the economy isn’t overheating. This is pretty much what we mean when we talk about “embedded” inflation.
Now suppose that policymakers want to bring inflation down. They have a problem: inflation has a lot of inertia. To get it down they need to give sellers a reason not to raise prices as much as they have been in the recent past. They can do this by pushing the economy into a recession. And if the recession is deep and long enough, the economy can be purged of inflation: not only will sellers stop raising prices as quickly, but they’ll begin expecting lower inflation in the future, which means smaller price increases, and so on. Eventually the economy can be reflated, at a permanently lower rate of inflation.
That is, however, a hugely expensive process, as we saw in the 1980s. So you really don’t want to let inflation get embedded in the first place. But how do you know if that’s happening?
The concept of core inflation goes back to a 1975 paper by Robert Gordon, who wanted to distinguish between “hard-core” inflation and what he called “bubbles,” but what I think are better described as “blips.” He offered a rough-and-ready solution: exclude food and energy prices, which fluctuate far more than the overall price level.
This approach has been hugely successful. It has been especially useful since 2007, when we experienced two inflation blips, in 2008 and again in 2011-12, that had many people screaming about a debased dollar and all that. The Fed stuck to its guns, asserting that the stability of core inflation showed that things were under control. And the Fed was right:
It’s important to understand, however, that the usual measure of core inflation is, as I said, just a rough-and-ready way to get at the difference between inertial inflation and short-term blips. The real distinction should be between prices that are sticky and prices that aren’t. And while just excluding food and energy has been a good approximation to that distinction in the past, past results might be no guarantee of future performance.
By the way, this isn’t news to economists who actually track inflation. In fact, the Atlanta Fed regularly produces a “Sticky Price Consumer Price Index” that tries to sort out goods and services by how frequently their prices change. I’ll be looking at that index a lot in the months ahead. Unfortunately, we don’t know whether it’s better than the standard measure of core inflation, because all the measures worked really well after the last crisis.
Why do I suspect that this time will be different? Mainly because the pandemic had weird economic effects, sharply depressing some activities while boosting others. And this probably means that we’re going to have a weird recovery too, with huge surges in things like travel, plus an unusual set of bottlenecks, like the global container shortage, resulting from the pandemic hangover.
So I expect to see a lot of price blips outside food and energy — some resulting from “base effects,” that is, recovery of prices that were depressed during the worst of the pandemic, some resulting from those bottlenecks. You can already see some of that in the latest consumer price report. For example, ordinary rents rose only 0.1 percent in March, but lodging away from home rose 6.6 percent.
What this suggests is that in the months ahead the numbers won’t speak for themselves. Headline inflation will surely be a poor guide to what’s really happening, even worse than it was in 2011-12. Even core inflation as usually measured may be misleading.
This doesn’t mean that we should discount inflation risks entirely. It does mean that we’ll need to kick the tires on whatever inflation readings we get, and try, as objectively as possible, to figure out whether or not they’re actually reason for concern.