Unemployment and Inflation: Can We Stop Using the Phillips Curve Already?

When a theory has been disproven, it typically gets discarded into the dustbin of history. Disprove that the Earth is at the center of the universe? You change your calculations accordingly. That should not just apply to science, but also to economics. However, some things die hard. As I was reminded by reading this recent article from the Council on Foreign Affairs and this November 2017 article from the Economist, the Phillips Curve is such an example.

What is the Phillips Curve? The Phillips Curve is an economic concept developed by A.W. Phillips that suggests that there is an inverse relationship between unemployment and wage inflation (see below). If unemployment goes down, wage inflation increases, and vice versa. Phillips conjectured that as the unemployment rate decreased, firms would be more likely to raise wages to attract scarce labor. Using the theory behind the Phillips curve means that fiscal or monetary policy could be used to adjust the economy to desirable labor market conditions.

In 1958, Phillips published his work on the unemployment and wage inflation rates in the U.K. from 1861 to 1957. This work led many economists to believe that there is indeed a solid, inverse relationship between unemployment and wage inflation. Although Phillips' work looked at wage inflation, economists applied it to general price inflation, which makes sense given that the prices that companies charge are closely related to wages. This economic framework is still believed to this day. Essentially, by creating inflation rate targets, policymakers now had a menu of contractionary and expansionary policy to adjust either for lower unemployment or lower inflation. Central banks across the world have used this logic to understand, forecast, and attempt to control inflation.

Something came along in the 1970s to disprove the Phillips curve: stagflation. Stagflation is when both unemployment and inflation get high. According to the assumptions of the Phillips Curve, stagflation should not have occurred, but that is precisely what happened in the United States during the 1970s. Looking at the Phillips curve from 1961 to 2015, we see a very tenuous inverse relationship between inflation and unemployment (see below). This relationship is even worse when considering that the economy performed better in low-inflation years than in high-inflation years. I know that correlations are typically not 100 percent, but the fact that there is no causation (nothing to say about correlation) should have invalidated the usage of this Curve in U.S. monetary policy.

While the simplified version of the Phillips Curve was disproven with the actual existence of stagflation, there have been a few alternative theories as to why the Phillips Curve isn't working "the way it should," modified Phillips curves, as well as analyses that include both short-term and long-term Phillips Curves (see below). Even Nobel Prize winning economist Milton Friedman argued that in the long-run, there is no correlation because employees will realize that their wealth increased nominally, and not in terms of real wealth (also see here and here).

I even have to question the short-run Phillips Curve. Why? The Director of the Philadelphia Federal Reserve co-authored a paper that was released last August (Dotsey et al., 2017). The conclusion of the Philadelphia Fed? Using the Phillips curve does not help predict inflation. At best, it might be somewhat helpful during an economic downturn, but even the authors concede that that assertion is far from conclusive (Dotsey et al., p. 27-28). The Minnesota Federal Reserve came to a similar conclusion in 2001 (Atkeson and Ohanian, 2001). The Federal Reserve Bank of San Francisco asked just this past October if the Phillips curve is flattening, and it doesn't look so good for the Phillips. It's not just in the United States where it has been put into question, but also in Europe. When looking at the minutes from the European Central Bank's July 2017 meeting, there is discussion on the disconnect between inflation and unemployment.

The U.S. Federal Reserve Bank's dual mandate is based on the assumption that the Phillips Curve is valid. The fact that the United States bases its monetary policy on a correlation that does not consistently exist should bother us. Right now, we have low unemployment and low inflation with a flat Phillips curve. Nevertheless, the Federal Reserve is making decisions based on a disproven economic theory. Instead of using invalidated economic model, how about the Fed tries making its decisions based on observable evidence?


3-9-2018 Addendum: More evidence that the Phillips curve doesn't work: this time, from the International Monetary Fund (Piazza, 2018).