Knzn asks a good question: Why, in my post on output gaps and inflation, didn’t I enforce a zero constant term — that is, no change in inflation when the output gap is zero.

The answer is, there are output gaps and then there are output gaps.

Our theoretical concept of an output gap is that it’s *defined* by the implications for inflation: potential output is the level of real GDP at which inflation neither rises nor falls, and the output gap
is the difference between actual and potential real GDP. Some estimates of potential GDP, like the CBO’s, are, in fact, constructed based on that concept — basically they’re based on an inverted
Phillips curve. And if I’d been using output gaps based on such an estimate, Knzn’s point would be exactly right.

In fact, however, I used estimates from the IMF, which gets its potential output numbers from a Hodrick-Prescott filter — basically, a trend estimate designed to smooth out short-term wiggles. I used the IMF numbers for two reasons: (a) laziness — the WEO database is very convenient for quick-and-dirty number crunching (b) avoiding circular reasoning. If potential output is estimated by assuming that inflation rises when output is above potential, finding that inflation rises when output is above potential … you get the picture.

For the US, these different methods don’t yield very different results. In other cases, they do — the IMF method interprets Japan’s lost decade mainly as a slowdown in potential growth. But using the IMF output gap numbers, there’s no necessary reason the intercept has to be zero.

In fact, that negative intercept is more or less predictable: the H-P filter forces a zero average output gap over the sample period, but inflation has in fact declined since 1980, so the results show that inflation declines when output is at “potential”.

Anyway, in the end it’s no big deal. The risk of deflation, on the other hand, is a very big deal.

Comments are no longer being accepted.