And Paul is right to be annoyed. Ever since this business started--even before Wesley Clair Mitchell--peak-to-peak measures have always been better than trough-to-trough or midcycle-to-midcycle ones.
A Fit of Peaks/a>: A Fit of Peaks Ah. I see that Greg Mankiw is pretending to be stupid in reacting to my post on revenue trends. Greg, of course, knows perfectly well why I chose the dates I did. But readers may not have gotten it, even though I explained it right there, Tolstoy and all.
So here’s the point: the biggest single influence on revenue is the state of the business cycle (which is why austerity in the face of a liquidity trap is so ineffective even at reducing deficits). And in normal times the business cycle depends much more on what the Fed does than on anything the rest of the government does (again, it’s different when there’s a liquidity trap). So what you want is to somehow abstract from the business cycle. And the easiest way to do that is to compare business cycle peaks, times when the economy is at more or less full employment. True, employment is fuller at some peaks than at others, but those differences are small, whereas the depth of the slump at recession troughs is much more variable. Hence, Anna Karenina.
The peak-to-peak interpolation thing is standard practice in many analyses; it’s how the Fed estimates capacity utilization, it’s how the Social Security Administration makes long-term projections, and so I’ve used it for revenues.
Those business cycle peaks don’t correspond to presidential terms. So? And here’s the thing: Reagan’s miracle must have been a very puny thing if it did nothing that shows at all when you compare the Reagan-Bush-era business cycle peak with the Carter peak.