Has the Phillips curve vanished? There is widespread discussion of the possibility that the Phillips curve has become horizontal. I am old enough to remember when all serious economists agreed that, in the long run, it is vertical. The reason for the thought that it might be horizontal is that large important countries have unemployment lower than estimated non accelerating inflation rates, and yet still have low inflation (including low wage inflation).
I looked at the old Europe 15 (the countries which were in the European Union in 1997 so the European Commission has harmonized time series going back to the 1960s). The Phillips curve is still very clear in that data set. There is no sign that it has a lower slope in the 2st than in the late 20th century. So I wonder why people are convinced it has vanished. One possible explanation is that more attention is focused on the USA than on other countries (maybe than on all other countries put together). So I will look at the case of the Phillips curve in the USA (probably the most over studied topic in economics). It still slopes down. Even in the US I estimate, at most, a mild reduction of the slope. This makes the puzzle more puzzling.

Before going on, I should note that I just learned that Joe Gagnon has written a much better version of this post right here .He argues that there is no inflation puzzle.

I will present the US evidence after the jump.

I will puzzle over perceptions here.

I will puzzle over perceptions here.

I think one key issue is that inflation hawks must argue that the Phillips curve is steep — not just downward sloping. It is hard to argue that we must accept 6% unemployment because 4.1% might lead to inflation of 2.5% for a few years starting next year. This means that a gradual increase in inflation (consistent with data from the 20th century) is considered anomalous because it contradicts the inflation hawk and generally austerian ideology. I think this is an important factor, but I have trouble determining how important using objective evidence.

second, Policy makers, commentators and even academic economists discussing real time economic issues focus intensely on the past months or, at most, the past few years. Reality is stochastic and the Phillips curve is useful but crude. We can’t really expect to be able to forecast the change in inflation from one month to the next. But we do. This means that there are constantly puzzles which turn out to have no interesting explanation. Gagnon puts this very well

The unexpected drop in inflation in mid-2017 is not particularly large in historical context. The figure displays the location of the latest observation (2017Q3), which is well within the historical Phillips curve experience.[5] In February 2017, the unemployment rate was equal to the estimated natural rate of 4.7 percent. By October 2017, it had dropped to 4.1 percent, implying that overall employment was 0.6 percentage points above potential. Based on the [accelerationist] Phillips curve, we would expect inflation to rise by about half a percentage point over the next year. But historical experience suggests that inflation may end up anywhere from 1 percentage point lower to 3 percentage points higher. In other words, the Phillips curve remains an important fundamental driver of inflation, but we should not overstate the precision with which it operates.

Third, I think there is great faith, based on little evidence , that the non accelerating rate of inflation (NAIRU is known and fairly high — definitely higher than it used to be back in the good old days (note that Gagnon does not share this view 4.7% is about what the NAIRU was guessed to be when the term was coined) . I think a very large part of this is the conviction that unemployment can not stay far above the natural rate for years and years. This is, I think, an article of faith. If unemployment can’t be far above the natural rate for years, fact that unemployment is much lower than it was in recent years implies it must be far below the natural rate. Marco Fioramanti and I have written a lot about this. I just include a google search.