Why Inflation Stays Low Even Though Unemployment Is Down

The Phillips Curve isn’t dead. You just need to know where to look for it.

To long-time baseball fans, the Phillips Curve sounds like a tribute to Lefty Phillips, the Los Angeles Dodgers pitching coach who helped Hall of Famers Don Drysdale and Sandy Koufax win National League pennants and a World Series title in the 1960s. 

The Phillips Curve is actually named after A.W. Phillips, a New Zealand-born economist who in 1958 noticed that wage growth is strong when the jobless rate is low. (Yes, kind of obvious to the layman.) American economist Paul Samuelson later made a bolder restatement of the Phillips Curve, saying that overall inflation—not just wage growth—is higher when labor markets are tight.

Except the Phillips Curve hasn’t behaved lately. The unemployment rate in November was just 4.1 percent, which should have been enough to set off a bidding war for talent, driving up wages and the overall price level. Yet inflation in November was just 1.8 percent, according to the measure that the Federal Reserve tracks, the yearly change in the price index for personal consumption expenditures. Stripping out volatile food and energy prices, inflation was just 1.5 percent. Both figures are below the Fed’s target inflation rate of 2 percent.

Economists have struggled to explain this mystery. Some have simply declared the Phillips Curve dead. But that seems extreme. Everything we know about markets says that when a commodity such as labor is in short supply, its price should go up.

One partial explanation is that the unemployment rate isn’t a great measure of the tightness of the labor market. True, the number of people who are unemployed and actively seeking work has shrunk. But the pool of unemployed is not the only source of new hires. As this chart shows, most people who start working in any given month weren’t even counted as unemployed the month before. They might have been in school or the armed forces or prison or just on the sidelines, not actively seeking a job. This source of new hires has become increasingly important in recent years as the unemployment rate has fallen.

Data: Bureau of Labor Statistics

The question then becomes, how many more people are still outside the labor force but available for work? The best indicator of that is the employment-to-population ratio. It’s simply the the number of people who are working as a share of the entire population. (Benn Steil and Benjamin Della Rocca wrote about this for the Council on Foreign Relations recently.)

As the following chart shows, the employment-to-population ratio fell sharply in the 2007-09 recession. It has begun to spring back but remains below its pre-recession peak, indicating that slack remains. Granted, the ratio is unlikely to get all the way back to where it was in 2007 because the Baby Boom is aging: a smaller share of the U.S. population is of traditional working age. It’s also true that some people who lost jobs are never getting them back because they’ve lost skills or motivation or contacts in the working world. But the employment-to-population ratio can probably get higher than it is now.


A skeptic might ask at this point whether the Phillips Curve inflation tradeoff works any better for the employment-to-population ratio than it does for the unemployment rate. In other words, do we really see inflation rising when the ratio goes up, indicating fewer people are available for work? The answer is no at first glance, but yes on closer inspection.

I went back to 1961 and looked at the correlation between inflation and the E/P ratio. The relationship was slightly negative, which seems to refute what I’ve been arguing. But then I looked separately at each business cycle and contraction over the period. While there were variations, by weighting the expansions and contractions by their duration I found that on average prices did tend to rise faster when employment was rising as a share of the population and to rise more slowly when employment was falling as a share of the population. This table sums up:

Data: Bureau of Labor Statistics

I suspect that the negative result over the entire period is the result of long-term trends that have nothing to do with the Phillips Curve. More women have entered the workforce since the early 1960s, which has raised the employment-to-population ratio. Meanwhile, the inflation rate has trended downward since the late 1970s because of better Federal Reserve policies, among other factors. Over the entire period, those two powerful factors mask the Phillips Curve. But the curve still shows up over shorter periods. 

Lefty Phillips could have told us as much: The curve is still useful even if it doesn’t always go over the plate.

    Peter Coy
    Bloomberg Businessweek Columnist
    Peter Coy is the economics editor for Bloomberg Businessweek and covers a wide range of economic issues. He also holds the position of senior writer. Coy joined the magazine in December 1989 as telecommunications editor, then became technology editor in October 1992 and held that position until joining the economics staff. He came to BusinessWeek from the Associated Press in New York, where he had served as a business news writer since 1985.
    Before it's here, it's on the Bloomberg Terminal.