Thoughts of the Week

Created: 2000-02-27
Last Modified: 2000-02-27
Go to
Brad De Long's Home Page

Teaching | Writing | Career | Politics | Book Reviews | Information Economy | Economists | Multimedia | Students | Fine Print | Other | My Jobs

Brad DeLong's mailing list | Comments on this page | Questions and Answers

What Happened to the Phillips Curve?

J. Bradford DeLong

February 2000

from The New York Times, March 9, 2000...

You remember the Phillips Curve, the relationship between unemployment and inflation proposed by British economist A.W. Phillips late in the 1950s and developed into its present form in the late 1960s by Ned Phelps and Milton Friedman. For more than 25 years mainstream economists' forecasts have rested on the idea that should unemployment fall below an unknown (but very real) level called the natural rate of unemployment, then inflation will start to rise. And inflation will keep on rising further until unemployment climbs back up to or above its "natural rate."

Today, however, the Phillips Curve is missing. All of a sudden economists are a lot less useful as forecasters. For as it stands now, economists' forecasts of economic growth, unemployment and inflation are no longer projections based on historical patterns but instead pure guesses based on gut feelings about when, how, and if the Phillips Curve will return. If they are to be of any use for forecasting, they need to come up with a better--and probably more complex--theory of under what conditions inflation is likely to rise.

When last seen during the 1990-1992 recession, the Phillips Curve looked very healthy. In the second half of the 1980's the unemployment rate fell to a low of 5.3% in 1989 and inflation, which had averaged 2.4% per year over 1984-1987, picked up to 4.1% per year and rising by early 1990.

Back then the Federal Open Market Committee--under the leadership of Alan Greenspan, who was relatively new as chairman at the time--reacted sharply by pushing up interest rates. Tighter monetary policy and the shock to expectations from the Iraqi invasion of Kuwait caused total spending to fall. Unemployment rose above its natural rate to a peak of 7.6% in June 1992. And inflation fell.

No one at the end of the 1980's knew exactly what the natural rate of unemployment was. But even the most optimistic did not think it could be lower than 5.5 percent and even the most pessimistic knew it could be no higher than 6.5 percent. But the end of the 1980's retaught the lesson that the Phillips Curve was very real, and had to be taken into account.

That's why officials in the Clinton Administration and the Federal
Reserve, as well as independent analysts, bit their fingernails throughout the 1990's as they awaited the return of rising inflation. By 1994 unemployment - at 6.1% - was in the range where inflation had started to accelerate in the late 1980's, so the Fed spoke of undertaking "preemptive strikes" against inflation.

Yet no inflation followed. By 1996 the unemployment rate was as low as it had ever gotten in the 1980's. Yet inflation fell to less than two percent a year. By 1999 unemployment was 4.2%--well below anyone's previous estimate of the natural rate--yet inflation was an even-lower 1.5%.

Where was the Phillips Curve? Until the end of 1997 there was confidence that the Phillips Curve would soon return. Temporary special factors--health care costs, rapid falls in computer prices, and so on--were momentarily retarding the tendency of inflation to rise when unemployment was lower than its natural rate. Such factors couldn't last forever, could they? Of course not. So when their influence came to an end, inflation would begin its rise. But years passed, and inflation did not rise.

So mainstream economists' opinion shifted to the belief that the natural rate of unemployment had fallen, though how far no one really knew. Economists began spinning theories of what had caused the natural rate to fall. Harvard's Jim Medoff began arguing in the early 1990's that technological and organizational changes had led the labor market to do a better job of matching workers needing jobs to vacancies, thus substantially lowering the natural rate. Others pointed to faster productivity growth that allowed higher wage increases to be consistent with relative price stability. Still others pointed to workers' fears for their jobs generated by the memory of the deep recession and high
unemployment of 1981-1983.

At some primal level, all economists still believe in something like a Phillips Curve. All believe that unemployment will fall if demand expands faster than the economy's long-run productive capacity. And all believe that if demand keeps on expanding faster than the economy's long-run productive capacity then, in the long run, inflation will rise.

It was just that the natural rate of unemployment--the signal that
this long run had arrived--had fallen mysteriously far and mysteriously fast.

But, truth be told, the Phillips Curve has not worked well outside
America. Economists Doug Staiger, Mark Watson, and Jim Stock pointed out in the _Journal of Economic Perspectives_ that even in the United States the Phillips Curve relationship was never as strong or as good at forecasting inflation as was taught in intermediate macroeconomics. And only in the United States has there been a relatively stable natural rate of unemployment to serve as a reliable indicator of when demand pressure is about to raise inflation. Elsewhere the causes of rising inflation have
always been too complex to be summarized by simply comparing unemployment to even a semi-stable "natural rate."

Thus perhaps the surprising thing is not that Phillips Curve-based
forecasts of inflation have gone awry in the past half decade. Perhaps the surprising thing is that the complicated economic processes determining changes in inflation could be summarized for so long by such a simple relationship as the standard Phillips Curve. In any event one thing is very clear: the simple theory of the relation between inflation and unemployment that economists have peddled for a quarter century no longer works; if economists are to be of any use, they need to come up with a better - and in all likelihood more sophisticated - approach to understanding why inflation rises.

Sign up for Brad Delong's (general) mailing list

Go to related links...

Add a comment on this page...

Dear Professor DeLong:

I am an economist at Goldman Sachs and read with interest your piece on the macro implications of the new economy, in particular the part about the NAIRU. I completely agree that faster productivity growth -- rather than changes in the labor market per se -- are the most plausible explanation for the apparently lower NAIRU. A few months ago I wrote a piece for Goldman along those lines, which I attach below. If you are aware of any references to (academic or other) work on this issue, I would be grateful if you could pass them along. Despite its apparent simplicity, I haven't seen anything on this idea apart from your notes.

I also noted with interest the comment on your piece (by a George Gelauff). The idea that productivity growth cannot affect the long-run NAIRU -- supposedly because this implies a further and further fall in the share of labor compensation in GDP -- is faulty but amazingly widespread. I have come across this view repeatedly in talking about the NAIRU with both clients and government economists.

Contributed by Jan Hatzius ( on September 1, 2000.

I really like your web site, just discovered today. I also liked your Phillips curve article. When I was in school there were several key points on which I disagreed with the conventional wisdom, and where developments in the years since to have proved my viewpoint to have been the correct one. Thanks to your article, I am pleased to be able to add my Phillips Curve interpretation to that list.

I argued 30 years ago (with absolutely no success in convincing my economics professors!) that even if the Phillips Curve was as fundamental an economic relationship as it was presented to be -- which I doubted -- that it did not follow that such an economic "blunt instrument" should serve as the basis for policy making. I argued then, and still believe, that other more benign economic control mechanisms should be used to achieve desired economic results, leaving the Phillips Curve effect (with its concommitant adverse impacts on jobs) to be only the policy weapon of last resort.

Thanks for your innovative thinking.

Contributed by Peter Eirich ( on March 30, 2000.

Dear Mr. DeLong:

What surprises me about the Phillips curve is that anyone ever took it seriously in the first place.

Back when inflation was "a problem," I used to study the various theories of the causes of inflation. In the case of all of them, I sooner of later found the point where the theory went " mumble, mumble, the money supply expands relative to the volume of goods and services produced." If the authors of the theories were pressed, they invariably explained that of course, if the money supply stayed stable, all kinds of terrible things would happen, but prices would remain stable. If they asked what would happen if the money supply was expanded without any of the things taking place that their theories used to explain inflation, they would claim it wouldn't happen, but eventually get around to admitting that prices would rise.

So expansion of the money supply relative to the supply of goods and services is necessary and sufficient to explain inflation. What therefore is the question the Phillips curve and other such things are

Contributed by Stephen St. Onge ( on March 17, 2000.


Thanks for your interesting commentary on the Philips curve. I was, however, surprised that you did not mention globalization and the rise of the NICs (and especially China) as a key factor conditioning price changes and aggregate supply dynamics.


Contributed by Barry Feldman ( on March 17, 2000.

Dear Prof. Delong,

I read your article in the International Herald Tribune and found it very interesting. In my work on portuguese stability programme the issue of growth potential and Nairu was raised by Commission officials. We agree to disagree. It does not mean that we are right - that is raising growth prospects and low inflation - or that a more conservative view is not better. In OECD forecasting business, the US case as you presented has been in discussion for some years - known as the hard-landing vs smooth landing scenario. After two or three years of dire predictions on the US economy, OECD just stop worrying about the latter. I agree with you that we need better theory to understand the US experience.

Yours sincerely

Contributed by Fernando Chau ( on March 17, 2000.

As long as the U.S. has a floating exchange rate, it is not clear to me that foreign competition puts that much downward pressure on inflation: the exchange rate adjusts, after all...

And the way I think of it is that you need the Phillips curve so you can figure out what is the appropriate rate of growth of the money stock...

Contributed by Brad DeLong ( on March 17, 2000.

Professor of Economics J. Bradford DeLong, 601 Evans Hall, #3880
University of California at Berkeley
Berkeley, CA 94720-3880
(510) 643-4027 phone (510) 642-6615 fax

This document:

Search This Website