Jared Bernstein on The Phillips Curve: The Fed, Jobs, interest rates, and economic models | Blog#42

The Phillips Curve is at best in ill repair, if not just outright broken.

Jared Bernstein
August 23rd, 2015


Here’s an excellent piece by Ben Leubsdorf on how the Phillip’s curve–long a workhorse of macro-policy–is not much help these days in trying to quantify the relationship between slack and inflation.

From the piece:

In an influential 1958 paper, [A. W. Phillips (a Kiwi, btw!–JB)] hypothesized that employers will bid up wages when workers are scarce, but feel little pressure to raise pay when unemployment is high and many workers are available. Data on the U.K. economy from 1861 to 1957 backed him up: High unemployment generally corresponded with low or negative wage growth; low unemployment was reflected in faster wage growth.

The sloped line depicting that relationship was embraced by many economists in the following decades to help explain how the economy works. Increasingly sophisticated versions of the so-called Phillips curve use measures of economic slack, model prices as well as wages and incorporate concepts like inflation expectations…

[Fed chair Janet]…Yellen has been a longtime booster, citing it as a Fed governor during policy discussions in the 1990s and telling lawmakers in 2010 that despite some shortcomings, “the Phillips curve model provides a coherent and useful framework for thinking about the influence of monetary policy on inflation.”

Yes, slack and inflation are correlated, but the extent of that correlation changes significantly over time, and lots of other factors get in the way–globalization (which increases both the supply of goods and labor, affecting both prices and wages), bargaining power, and the extent to which the Fed itself “anchors” inflationary expectations.

As the forthright David Altig from the Atlanta Federal Reserve puts it: “We haven’t lost faith in the framework [but] the numbers that you would plug into that framework and the exact levels at which the pressures begin to emerge, we’re not so clear on those.”

I myself, riffing off of work by Ball and Mazumder, have shown how the slope of the curve–the magnitude by which diminished slack drives up inflation and vice versa–has drifted about over time, landing around zero in recent years (see the figure here).

But this figure from the Leubsdorf piece tells the story perfectly well.


That doesn’t look like a lamppost that could shed much useful light on the Fed’s dual mandate (maintain full employment at stable inflation). Which makes this a timely discussion, because this is the week the Fed holds its annual retreat in Jackson Hole, Wy. I can only hope someone throws the figure above up on the screen, and says to their colleagues, “Really? Seriously?? This is the heart of our model?”

There are at least two responses to the questions this raises. The correct ones are “what’s wrong with the model? What factors are changing this historical relationship and how lasting are they? And what does this imply for monetary policy, especially our imminent interest raising campaign?”

Then there’s the response from Dennis Lockhart, president of the Atlanta Fed:

“In the absence of direct evidence that inflation is in fact converging to the target and in the absence of compelling or convincing direct evidence, I think a policy maker has to act on the view that the basic relationship in the Phillips curve between inflation and employment will assert itself in a reasonable period of time as the economy tightens up.”

To me, that sounds downright religious, not at all empirical, and certainly not in the spirit of what we’ve been told is a data-driven Fed.

Blogger’s Note:

Aren’t rules, regulations, laws in any functioning society all based on the assumption that most will not game them? How can that be any different with economic models and prescriptions?

In the end, we can’t have real economic discussions about the symptoms of this or that economic factor or phenomenon and how to fix it until we solve the political end of the problem. That, in my estimation, is about 90% political in nature.