In general this is a good rule. The data are highly erratic and subject to very large revisions. I once bet a friend that there was no statistically significant relationship between the final data and the numbers originally reported. He won, just barely, because we agreed to accept a 10 percent significance level. Of course, if we had the comprehensive GDP revisions available at the time time, I think that I would have won. Anyhow, this NYT article is especially off base for presenting comments about the strong productivity performance reported for the second quarter. Productivity growth is especially erratic around downturns. In the first quarter of 81 it grew at a 6.6 percent rate (just before the onset of the 1981 recession). In the second quarter of 91, it grew at a 6.1 percent rate (just after the recession officially ended). In the second quarter of 01 productivity grew at a 7.4 percent rate and the 4th quarter at a 5.8 percent rate. The reason the numbers are erratic is that the recorded rate of productivity growth has everything to do with the timing of layoffs. If businesses wait a quarter or two and keep workers on the payroll who aren't doing anything, then productivity looks bad over those quarters. When they do get laid off, productivity suddenly looks great. The moral of this story is there is no longer term takeaway about yesterday's data on productivity growth.
--Dean Baker
