Arnold Kling  

Random Notes on the Phillips Curve

Price Controls and Black Marke... Greens, The Road, and <...

Jeffrey C. Fuhrer wrote in a 1995 article entitled, "The Phillips Curve is Alive and Well."

Overall, then, conventional tests of the stability of the Phillips curve indicate remarkable stability. There may be no other macroeconomic relationship that could perform as well by these criteria.

Fuhrer defines the Phillips Curve as an equation to predict quarterly inflation based on lagged values of inflation and lagged values of the unemployment rate. This version of the Phillips Curve has tremendous momentum (when inflation is high, it tends to stay high), with a slight nudge from the lagged unemployment rate.

Note that the stability that Fuhrer finds is between a pre-1980 Phillips Curve and post-1979 data. In contrast, the instability that traumatized those of us who lived through it was between a pre-1970 Phillips Curve and the post-1969 data.

As I pointed out four weeks ago, the heyday for the Phillips Curve was 1956 through 1969. One way to characterize this period is that we had a stable Misery Index, defined as the sum of the annual averages of inflation and unemployment

From 1955 through 1968, the MIsery Index ranged from 6.36 to 8.60. Over this period, inflation ranged from 0.67 to 4.71, and unemployment ranged from 3.56 to 6.84. So each individual series had much wider variation than the sum of the two.

Then came the 1970's: the Misery Index ranged from a low of 9.01 in 1972 (Nixon's landslide re-election came with a falling unemployment rate and inflation held in check by wage-price controls) to a high of 19.53 in 1980 (President Carter's last year in office). Note that there was no overlap between the pre-1970 Misery Index and the decade that followed.

Fuhrer's equation has a NAIRU of 6 percent, which means that inflation is being nudged down when unemployment is above that, and inflation is being nudged up when unemployment is less than 6 percent. By that standard, inflation was being nudged up every year from 1961-1974, with particular strong pushes from 1966-1969, when unemployment was below 4 percent. One could argue that Presidents Ford and Carter paid the price for past attempts to keep unemployment too low.

Perhaps the most awkward period for the Phillips Curve was 1977 through 1980, when in spite of the fact that unemployment was over 6 percent, the inflation rate accelerated sharply.

The next most awkward period was 1997-2000 (after Fuhrer's article), when the unemployment rate stayed below 5 percent. We should have reaped a whirlwind of inflation from that, but we did not. Of course, my story for the 2001-2003 period is that the labor market was actually very weak, but lots of folks dropped out of the labor force, keeping the unemployment rate artificially low. If you believe that story, then you cannot blame the Fed for keeping interest rates below Taylor-Rule levels.

Looking ahead, the next 12 to 18 months should be interesting. The unemployment rate has been so far above 6 percent for so long that if the Fuhrer equation holds up, we should be seeing some pretty strong downward pressure on inflation. Instead, if inflation remains between 0 and 2 percent, this will look to me like another anomaly for the Phillips Curve.

Comments and Sharing


COMMENTS (4 to date)
magilson writes:

Who were these "lots of folks"?

Jacob Oost writes:

I'm not a professional economist, but the Phillips Curve (and the ensuing policies it entails) strike me as the economic equivalent of using leeches to balance humors and reading forehead bumps to predict what career I'll have.

Gary Rogers writes:

Inflation and unemployment behave as though they have some of the same causal variables, but that does not mean they will always move together. For example, uncertainty or panic will generally cause unemployment to go up and inflation to go down as the Phillips Curve predicts. If the uncertainty is about expected inflation, though, it could cause higher unemployment due to the uncertainty while also yielding higher inflation due to the expectation. This would be contrary to the Phillips curve but still fit the expected results of the causal variables. An easy money supply will also encourage people to pull money out of savings and invest in projects that could also reduce unemployment. That is, unless the easy money leads to expected inflation, in which case people pull their money out of savings and invest in gold or other tangible assets that will act as a hedge against inflation. Things like this can and should be included in the models, but if the equation is simply that when inflation goes up unemployment goes down the model would certainly lead to some wrong assumptions. The problem with regressions is that they only show correlation over time and do not indicate cause and effect.

Floccina writes:

Didn't we learn that it is the change in the rate of inflation that matters rather than the rate itself?

Comments for this entry have been closed
Return to top