Phillips Curve

In Arthur Laffer’s case, it might have been carelessness in the use of words, but the latest attack on the Phillips curve by Larry Kudlow (discussed by Mark Thoma and pgl) is unmistakably either disingenuous or ignorant. (I'd prefer not to use such inflammatory terms, but I just can't think of a nice way to say it.) I single out Larry Kudlow only because his is the most recent example. I’ve heard similar arguments in the past, and I’m never sure whether the people making the argument realize that they are attacking a straw man – a man made completely of straw, that is, with no hope of even finding the Yellow Brick Road. But it’s one or the other: either Larry Kudlow is trying to pull the wool over people’s eyes, or he is somehow unaware of the last 40 years of Phillips curve research.

His argument essentially goes like this: the Phillips curve is an inverse relationship between unemployment and inflation. What we observe, however, is that the relationship is positive, not inverse. Therefore the Phillips curve is wrong.

That might be a strong argument against the version of the Phillips curve that was widely believed during the 1960s (the “groovey” Phillips curve, as Gabriel Mihalache called it in an earlier comment). That version posited a stable relationship between the unemployment rate and the level of inflation, the same relationship in the long run as in the short run. The problem is, nobody believes in that version of the Phillips curve any more. Don’t even bother looking in the duffle bag. That argument was settled conclusively before Bill Gates dropped out of college.

For practical purposes, the Phillips curve – the version that is used to guide forecasts and generate policy prescriptions today – is not a relationship between unemployment and the level of inflation; it is a relationship between unemployment and the change in the inflation rate. (There are subtle reasons why this practical definition isn’t exactly right, but for such a simple formulation, it comes awfully close. Econometrically speaking, the Phillips curves fit by people like empirical Phillips curve guru Robert Gordon have exactly this “accelerationist” property.)

So let’s look at the data. Suppose we don’t even bother with all the econometric subtleties that people like Robert Gordon and Mark Thoma would urge upon us. Suppose we just do a straight scatterplot. Here it is:



I don’t know about you, but that sure looks like an inverse relationship to me. Want to include the 1970s? OK.