Rogé Karma has a backgrounder on the history of the Phillips Curve. He does not use the term, but the idea was that there is a trade-off between unemployment (of labor) and inflation: low unemployment/high inflation and vice versa. This sets up a dilemma for the Fed which is supposed, by Congressional mandate, to seek stable prices and maximum employment. The locus classicus of the idea was [Here beginneth the potted history.] the period of stagflation in the’70’s terminating with high inflation that was broken only when Paul Volker had the “courage” to impose high interest rates, resulting a sharp recession and unemployment but with inflation reduced.
There was a reason to think there could be a trade-off between unemployment and inflation in the 1970’s. Was it the real reason? I don’t know, but the theory made sense. The idea was (simplifying as any model has to) that the macroeconomy consisted of one output – GDPStuff – produced with one input, Labor. As with any simple supply and demand analysis there is a price of Labor at which it was fully employed and GDP was maximized. In this scheme if we suppose that wages, the price of labor, was arbitrarily set higher than the equilibrium price (“union power!!!!”) the amount of labor demanded and the amount of GDPStuff produced would fall – recession and unemployment. A partial explanation for how this could happen is that firms would agree to higher wages because they expected to be able to raise their own prices and if the Fed validated this with expansive enough monetary policy, this would turn out to be a correct expectation across the whole economy. In effect the Fed would be inflating the price of GDPStuff so that the real, inflation-adjusted wage was reduced and unemployment and recession reversed or avoided.
Play this sequence out several times and the observed result would be the Phillips Curve tradeoff.
It’s important to note that without inflationary expectations this would not play out; there would be one spurt of inflation and full employment would be restored. Expectations (and an always accommodating Fed), however, will produce a “wage-price spiral.” But the idea was that if one knew how expectations were formed, the trade-off between inflation and unemployment would be stable and the Fed could at least choose the least bad combination of unemployment and inflation, what Arthur Okum of the University of Michigan economics department where I was in graduate school termed the “misery index.” [Oh, unhappy memories of 1967 trying to estimate a distributed lag inflation expectation equation when we had to solve regression equations on a mechanical calculator! (We did have piped water and indoor toilets, however. 😊)]
With inflation “tamed” and inflation expectations (and “union power?”) “broken” or “anchored” by a credible Fed inflation target, the macro economy was set for les trente glorieuses of low inflation. One could still think that a Philips Curve existed, just that we were on low inflation points along the curve. The Great Recession with a decade of persistent under-target inflation and glacially slowly falling unemployment was not inconsistent with the idea of a stable Phillip Curve.
Hence as Karma recounts, last year there was a “‘traditionalist view’ of monetary policy: the belief that the only way to tame inflation is by causing a recession. This view so thoroughly dominates the economics profession that it is often considered something closer to a law of nature. It is why, when inflation began taking off last year, nearly every economist, forecaster, and CEO believed a recession was around the corner.” [Last October Bloomberg reported a “consensus” of economists was the probability of a recession in 2023 was 100%.] The counter belief – the consensus was never 100% -- that a “soft landing” was possible was mocked with the phrase “immaculate deflation.”
So here we are in late 2023 with inflation having come down sharply with no recession, no unemployment. Is the Philips Curve pasé or one of Krugman’s “zombie ideas.” I think so and the reason is that the economy is a lot more complicated than GDPStuff produced by undifferentiated Labor. And it is complicated in a significantly different way. Wages are not the only relative price than could get out of equilibrium and in point of fact were NOT the point through which supply chain and the Putin oil shocks impacted the economy in 2021, shocks that the Fed very properly allowed inflation to facilitate a readjustment of relative prices when some prices move down with difficulty if at all. In this view, there is no one Phillips Curve in labor unemployment and inflation but a myriad of “Phillips Curves” in diverse markets.
This does not however make the Fed’s job easy. It does imply the possibility of reducing the inflation — inflation that the Fed itself allowed in order to facilitate adjustment to non-labor market shocks — without producing unemployment of labor. It does not guarantee that it will be able to do so. As the Fed tries to squeeze inflation from the system, it could go too far, reduce inflation too much, produce unemployment and a recession and voila; the Phillips Curve would be reborn.
Nice post.
Your story reminds me of something Robert Solow often pointed out - maybe he still does: that the Phillips Curve was originally observed as a relationship between unemployment and **wage** inflation (not the change in the overall price level). This creates exactly the possibility that you mention - that there could be many Phillips curves, one for each relative price, and therefore a lot of noise and information loss in generalizing these relationships (many of which might be zero correlation) into one between unemployment and a non-relative, aggregate-price-change headline number.
Contrast the humility that would result from such an awareness with the pose struck by our situation-room-dominating friend, Larry Summers, who can to all appearances sail these monetary policy seas with scarcely any reference to a model , and with no need for any kind of compass other than constantly repeating the phrase "history tells us..." It's an impressive performance - and a frightening one. It's as if Larry died and came back as his own private Phillips Curve. All he seems to care about is what unemployment rate was "required" over the last 50 or so years to reverse an episode of inflation. And he keeps insisting that inflation isn't **really** reversed until such and such an unemployment number is attained.
I find discussions on inflation confusing and contradictory. Like when the inflation was just getting started some economists were saying it was the start of stagflation and others saying it was transitory. Then when the Fed started hiking some said there would be a recession and others a soft landing. It was so contradictory I developed an unrealistic toy model to try to make some sense of inflation in the 1970's and early 1980's, with low inflation in the 2010's and inflation in 2022.
I put it together in Dec 2022 and wrote it up as my first substack post the next month.
https://mikealexander.substack.com/p/a-new-way-to-look-at-inflation-revised
I use a money balance on the real economy and use a correlation between that balance and historical NAIRU values to obtain an estimate for the trend in NAIRU over time. I then compare unemployment rate with the NAIRU to identify inflationary periods, where inflation manifests or the Fed is maintaining high rates to suppress inflation, and deflationary periods when one can have low unemployment, low interest rates, fiscal deficits *and* low inflation such as the period just before the pandemic:
Today we have unemployment running 0.9% below NAIRU while it was right around NAIRU in 2019, so we had no inflationary forces in 2019, but we have them now. So, the Fed is going to have to keep the interest rates high and even then, isn't going to get inflation back down to 2% without more hikes.
It appears that rate hikes have worked by suppressing job openings rather than unemployment since these have come down a lot (see Beveridge curve link) while unemployment has stayed low.
https://mikealexander.substack.com/p/the-beveridge-curve-and-cultural
Since deficits are likely going to go in the future, the model says so will NAIRU, implying the economy will get more inflationary. The Fed may have to hike further, and we may see job opening fall faster and maybe going into recession in 12-18 months? I don't know, I am trying to see if this model has any predictive power or whether it was just a fancy curve fitting exercise :)