The natural rate of unemployment is not something we hear a lot about in academic circles these days, and it's out of fashion even in some central banks. I was out of town when this happened, but when I was working for the St. Louis Fed, Larry Meyer showed up at the Fed to talk to economists in the Research Department. One of things he wanted to know was our estimate of the natural rate of unemployment. Meyer seemed offended, apparently, that we had never thought about it. He didn't know that it's hard to find anyone at the St. Louis Fed who would take the Phillips curve, let alone the natural rate of anything, seriously.

I was reading Paul Krugman's blog, and he is saying that recent evidence "seems to have brought skepticism about the natural rate to critical mass." That sounds promising, so I thought I would like to understand what Krugman is getting at.

Krugman thinks that the "natural rate hypothesis," which he attributes to Milton Friedman, was an influential idea that we should re-assess. So, what did Friedman actually have to say about this? If you read Friedman's "The Role of Monetary Policy," you'll find that the natural rate hypothesis is no more nor less than the long-run neutrality of money. Friedman argued that the central bank could control nominal quantities - the nominal quantity of central bank liabilties outstanding, the price level, inflation, for example - but that it would fail in any attempt to permanently control real magnitudes. To get that idea across, Friedman told a story. That is, there exists a natural rate of unemployment - roughly, the rate of unemployment that would exist in the long run in the absence of aggregate shocks - and if the central bank endeavors to force the unemployment rate to be higher or lower than the natural rate then this would lead to ever-decreasing or ever-increasing inflation, respectively (though in the ever-decreasing case, presumably there would be a lower bound due to the lower bound on the nominal interest rate). A key part of the story is a theory of the short-run nonneutrality of money. For Friedman, this is a theory of money surprises, relying on adaptive expectations. Workers supply labor based on the real wage they expect, so if higher-than-anticipated money growth causes growth in nominal wages and prices to exceed what is expected, then workers supply more labor, thinking their real wage is higher, and firms hire more labor, as they know that real wages have fallen. That's a theory of Phillips curve correlations. Money surprises cause output and inflation to move in the same direction.

Lucas constructed a closely related theory of money surprises and nonneutrality, and in the process introduced rational expectations to the macro profession. Lucas has a theory of Phillips curve correlations, and can also say something about how the Phillips curve shifts with the monetary policy rule. For example, the slope of the curve changes with the degree of noise in the policy rule. New Keynesian (NK) models can of course produce Phillips curve correlations as the result of sticky prices. Expectations are rational in such models (the baseline ones anyway), and there is no imperfect information about aggregate shocks, but monetary policy is not neutral because some prices are locked in from past decisions. Unanticipated monetary expansions then lower relative prices for firms which cannot change prices in the current period, and those firms increase output to meet demand.

In principle all these Phillips curve models - Friedman, Lucas, NK - can give rise to the process Friedman describes, i.e. a process by which monetary policy can mess things up when the central bank attempts to peg real quantities. In NK models this would require some work. But the idea might be to have pricing rules respond to observed central bank behavior. The central bank tries to peg the path for aggregate output, but then firms change their pricing behavior in response, and this leads to either increasing or decreasing inflation.

In any case, I cannot find any evidence in Friedman's work that he thought measuring the natural rate of unemployment would be a useful thing to do, or that shocks independent of monetary policy causing movements in the unemployment rate would necessarily lead to movements in the inflation rate in the opposite direction. So those ideas are coming from somewhere else, and they are well-entrenched in the thinking of central bankers. Paul Krugman believes this too, as he says:
I’d say that the preponderance of evidence still supports the notion that high unemployment depresses inflation, low unemployment fosters inflation.
That's a particularly poor way to think about inflation. As David Andolfatto likes to tell us, unemployment does not cause inflation. Some shocks to the economy cause unemployment and inflation to move in opposite directions, for example in the Spanish example that Krugman shows us in his post. In other cases - for example during most of the post-2009 period in the United States - inflation and unemployment move in the same direction, more often than not.

Further, the prevalent idea in central banks currently is that, once the unemployment rate falls below the natural rate, inflation will take off. This idea is built on a misunderstanding of Friedman's thought experiment in his 1968 paper. The results of a monetary policy experiment don't tell us how inflation responds to other types of shocks that could be moving the unemployment rate around.