Send any friend a story
As a subscriber, you have 10 gift articles to give each month. Anyone can read what you share.
By Paul Krugman
The other day, Bloomberg’s Tracy Alloway published an interesting article on the sudden popularity of a newish economics term: “immaculate disinflation,” meaning falling inflation without any rise in unemployment. She tried to trace down the origins of the term and found that many people credited … me.
What actually happened, in terms of the etymology, was that I picked up and popularized a phrase from the late John Williamson, who described some misguided views about international trade as “the doctrine of immaculate transfer.” Later, I and others adopted the same language for some views about inflation. Fun stuff, and few things flatter a writer as much as seeing a term he more or less coined get widely adopted.
But the more important point is the speed with which the conventional wisdom on inflation has shifted. Just a few months ago we were hearing many pronouncements to the effect that containing inflation would require years of very high unemployment. Now, suddenly, we’re hearing that controlling inflation may cost nothing at all.
I always considered extreme pessimism about disinflation to be sensationalist scaremongering; the current extreme optimism, on the other hand, makes me a bit nervous. So I thought some readers might be interested to read a potted history of the relevant economic doctrine with some actual data attached.
Warning: This is my “wonking out” newsletter, and it will contain not just charts but even (the horror) a few equations. If that makes you break out in hives, stop reading now.
OK, then. In the beginning was the Phillips curve, a purported downward-sloping relationship between unemployment and inflation. Such a relationship isn’t hard to justify. When the economy is running hot, you would expect to see rising prices: Firms believe they can get away with charging more, workers believe they can demand higher wages, and with wages and prices rising, producers will see their costs going up and raise prices further. A hot economy also tends to have a low unemployment rate, so some correlation between unemployment and inflation makes sense.
In 1958, the New Zealand-born economist A.W. Phillips found such a correlation between unemployment and wage growth in Britain, and the idea of an unemployment-inflation trade-off was adopted by many economists. (The 2 percent inflation target also has its origins in New Zealand. Are Kiwis the root of all economic evil?)
In the late 1960s, however, Milton Friedman and Edmund S. Phelps argued persuasively that inflation also depends on expectations of future inflation. The experience of the 1970s seemed to bear that out, and it became standard for economists to write down an “augmented” Phillips curve looking something like this:
Inflation = -b × (u-u*) + expected inflation
Where u* is what Friedman called the “natural” rate of unemployment, the rate at which actual inflation equals expected inflation, u is actual unemployment, and b is the rate at which wages and prices adjust to markets.
Now, only a silly economist (of which there are a few) writes down an equation like that in the belief that it represents truth. It’s more a matter of trying to keep yourself honest, representing your ideas with explicit algebra for the sake of clarity — and also making it easier to prove yourself wrong.
Still, let’s run with it for a while. Obviously, it matters what determines expected inflation. In the 1970s it seemed reasonable to assume that firms and workers based their expectations of inflation on the recent past. If you assume that expected future inflation is just last year’s inflation, the Phillips curve becomes:
Inflation = -b × (u-u*) + last year’s inflation
Which you can rearrange to get:
Change in inflation = -b × (u-u*)
And for a while this last equation actually seemed to work, sort of. The Congressional Budget Office produces an estimate of u*, which it calls the noncyclical rate of unemployment. And if you plot the difference between actual and noncyclical unemployment against the change in core inflation (excluding volatile food and energy prices), it looks like this from the ’60s through the mid-80s:
Physics it isn’t. But as you can see, high unemployment did seem to be associated with falling inflation, low unemployment with rising inflation.
Unfortunately for economists, although fortunately for the economy, that apparent relationship vanished in the 1990s. What we seemed to see instead was the re-emergence of the old simple correlation between unemployment and the level of inflation:
What happened? The conventional wisdom is that after a long period without episodes of major inflation, people stopped updating their inflation expectations based on the recent past. Surveys suggest that expected inflation didn’t just drop in the 1980s, it also became anchored, no longer changing much even when there were brief inflation spikes:
Then came the pandemic and the policy response to it. Suddenly inflation soared to levels not seen since the early 1980s. This shouldn’t have happened according to the standard Phillips curve, but there was a lot of weird stuff going on.
Indeed, things have been weird enough that it’s not at all clear what measure of inflation we should be using. For what it’s worth, here’s old-fashioned core inflation — using three-month changes rather than annual rates in an attempt to smooth out some of the statistical noise while keeping up with a rapidly changing situation:
But everyone working with the data these days knows that traditional core has become problematic in the plague years. The numbers have been buffeted by new sources of volatility, such as used car prices; the official measure of shelter prices, which mostly reflect rents but with a long lag, has been distorted by a huge rent surge in 2021-22 that ended months ago but is still filtering into the published numbers; and new problems keep emerging. For example, the White House Council of Economic Advisers, while producing some new, reassuring wage numbers, also cautions us that the Consumer Price Index has recently been understating health insurance costs.
So a lot of effort has been going into estimating how many angels can dance on the head of a pin — I mean, “true” underlying inflation. It’s a worthwhile, even important exercise. But the bigger picture is the speed with which inflation, however one imperfectly defines it, first soared, then plunged.
When inflation surged, several influential economists argued that we were back to the bad old days represented in my first chart, in which the harsh logic of the Phillips curve meant that reducing inflation would require years of high unemployment. I could say that I found the emergence of such views puzzling, because nothing in the available evidence pointed to a return to traditional stagflation, and the strangeness of our times had clearly complicated the traditional thinking on inflation.
But the truth is that I wasn’t just puzzled, I was angry — it seemed to me that declarations that the ’70s were back were sensationalist and irresponsible fearmongering.
As you might guess, I have views about what actually happened. But this newsletter has already gotten too long.
For now, my point is that we obviously aren’t rerunning the ’70s. While I’m cautious, on the other hand, about fully embracing the doctrine of immaculate disinflation, inflation does seem to be coming down as quickly and easily as it went up, and without too much economic pain.
Site Information Navigation
Source: Read Full Article