Economists care a lot about unemployment. And they care a lot about inflation. So much so, in fact, that for years they were really into this thing called the Phillips curve, which has unemployment on the X axis and (wait for it) inflation on the Y axis. Most of time, there's at least a short-term tradeoff between the two: If you let inflation go a bit higher, you could get lower unemployment, and vice versa.

But the experience of stagflation in the 1970s — where both unemployment and inflation were quite high — made a lot of economists skeptical of the relationship, though many people still find value in it, and it's made a bit of a comeback in revised "New Keynesian" form (see, for example, Galí and Gertler's version).

All that intellectual history is just prelude to the biggest development in Phillips curve research, which comes in a new-to-me paper by Gregor Smith at Queen's University, which looks at the curve for Japan between 1980 and 2005. Here's what happens when you plot the unemployment rate multiplied by -1 against inflation:

Now look at a map of Japan:

Haters may try to deny the realness of the relationship here with complaints like, "that's just a coincidence" and "why would you plot it against the unemployment rate multiplied by -1? why not just the unemployment rate?" These people are unwilling to see the clear empirical relationship before their eyes. More striking still, it appears that New Zealand's Phillips Curve ...

... looks not, you know, totally dissimilar to New Zealand.

France, by contrast, appears to defy the trend. Its Phillips curve ...

... doesn't really resemble the trademark pentagonal dimensions of Gaul.

More research is clearly needed into the geographic determinants of Phillips Curves. Thanks to Josh Freedman for alerting me to this important relationship.