TCS Daily

Is the Phillips Curve Dead or Alive?

By Kevin Hassett - June 26, 2000 12:00 AM

Alan Greenspan and his colleagues on the FOMC gather tomorrow to discuss the possibility of increasing interest rates for the seventh time in the past year. At the last meeting, when the Fed raised short-term rates a whopping half a percentage point, most analysts predicted that another dose of interest rate medicine was likely this time around as well. But now, most economists who watch the Fed closely believe that the rate hikes are over for a while now. Why the big change? The short answer is this: The Fed still relies on a topsy-turvy "old economy" model of inflation, and that model has been pleased by the latest twists and turns in the data. Most importantly, the unemployment rate rose in May from 3.9 percent to 4.1. The model that rejoices in such bad news is "the Phillips Curve," and you can't begin to think seriously about the new economy without looking at it closely.

In 1958, A.W. Phillips published one of the more influential articles in the history of economics. Phillips, an engineer by training, constructed a simple scatter plot of unemployment rates and the rate of change of wages using data from the U.K for the years 1861 to 1957. He saw something very striking. The data settled in fairly precisely on a single, downward-sloping curve. When unemployment was low, inflation was high. When unemployment was high, inflation was low. Most importantly, when unemployment was "just right" so to was inflation.

Economists find reliable relationships about as often as the Red Sox win the World Series, and the Phillips paper caused a great stir, a stir that continues to this day. If the Phillips curve is real then it provides Fed policymakers with a priceless guide. If unemployment is too low, inflation will be too high, and the Fed should tighten. What could be simpler?

Economists are generally loathe to attempt to exploit a relationship until they can think of a logical explanation for why it might exist, and, a huge amount of effort from our best researchers has gone into understanding the logical foundations of the Phillips curve. The best explanation for why the Phillips curve might exist works like this. If the Fed prints a little too much money, then the price of everything should rise a little bit. But suppose that workers have wages that are set by union contracts. Then wages don't respond immediately to the slightly higher inflation, so businesses actually get to pay workers the same wage while they get a higher price for their product. This is a good deal, so firms decide to hire more workers, which lowers unemployment.

Notice that this works only if you are able to fool workers for a little while. If they know that the extra little stimulus is coming, they will demand higher wages immediately, which will offset the price increase in the firm's ledger and cause unemployment to stay about the same. The fooling might not have to be too ridiculous however. If it is a royal pain to renegotiate contracts, it might make sense to agree to adjust things only once a year. In that case, short term monetary stimulus might still have a Phillips-curve effect. However, if all players in the game are sensible, one might expect the effects to be small. Firms might want to hire more workers when their prices increase because of a monetary stimulus, but they know that a year from now they will have to make good with higher wages.

If it is difficult to hire and train new workers, they might decide to not bother trying to take advantage of the short-term opportunity afforded by increased prices. New economy models, which we will look at in more detail in coming weeks, argue that the information technology advances we have seen lately have greased the gears of the economy. This increases the efficiency of labor markets, and softens the Phillips curve effects. So the Phillips curve might be an enduring relationship that the Fed can use to guide policy, and it might not. Which is it? We don't know for sure. As the following chart demonstrates, the Phillips curve was dead and buried until a few months ago. At the moment, a rotted claw has just emerged from the grave.

The chart plots the movements of the unemployment rate and the inflation rate in the U.S. for the years 1946 through May 2000. The Phillips curve relationship says that inflation and unemployment should be negatively correlated. When you plot variables that are negatively correlated, the plot should look like a pair of eyeglasses, with one series tracing out the bottom half and the other tracing out the top. In the first half of the sample, the chart looks exactly like that, the Phillips curve is in the data! When inflation goes up, unemployment goes down. When inflation goes down, unemployment goes up. Eureka!

But wait. Look what happens next. In the 1970s, the relationship is not nearly so nice, stagflation leads to periods where both of them go up together. And in the 1990s, which are blown up in the second chart, there seems to be almost nothing left.

Since 1993, unemployment drops like a rock, and inflation as of the May 2000 reading is almost exactly where it was 8 years ago. So the Phillips curve is dead, right? Not quite. Look at the last 12 months and you see something worrying. Just as we were about to write the obituary for the Phillips curve, the darn inflation rate started to climb up a little bit. The inflation reading at the end of the chart is the highest in four years. Maybe we have spent the last few years building up inflationary pressures, and the lid is about to blow off. Maybe the line at the end of the graph is going to the moon. That certainly is what has worried the Fed lately.

Which explains why the markets have been calmed by the increase in unemployment that we saw in the last unemployment report. If the new economy view is right, then the Fed should not be raising interest rates. The inflation rate is just blipping up a little at the end of that chart because data are noisy. The next move of the CPI will be down, and the Phillips curve will soon be put to death once and for all. On the other hand, if the old economy, Phillips curve view is right, then the higher unemployment rate suggests the Fed doesn't need to move. Higher unemployment will lead to lower inflation. Either way, the pressure is off, for now. So it will be a shocker if the Fed raises rates tomorrow, but we are at a tremendously interesting crossroads. The next few months of that chart will help us determine whether the Phillips curve is dead, undead, or alive and well.


TCS Daily Archives