What does the new gdp report imply for structural explanations of our current troubles?
How should we revise structural interpretations of unemployment in light of the new gdp revisions? (For summaries, here are a few economists’ reactions to the report.) Just to review briefly, I find the most plausible structural interpretations of the recent downturn to be based in the “we thought we were wealthier than we were” mechanism, leading to excess enthusiasm, excess leverage, and an eventual series of painful contractions, both AS and AD-driven, to correct the previous mistakes. I view this hypothesis as the intersection of Fischer Black, Hyman Minsky, and Michael Mandel.
A key result of the new numbers is that we had been overestimating productivity growth during a period when it actually was feeble. That is not only consistent with this structural view but it plays right into it: the high productivity growth of 2007-2009 now turns out to be an illusion and indeed the structural story all along was suggesting we all had illusions about the ongoing rate of productivity growth. As of even a mere few days ago, some of those illusions were still up and running (are they all gone now? I doubt it.)
On one specific, it is quite possible that the new numbers diminish the relevance of the zero marginal product (ZMP) worker story. The ZMP worker story tries to match the old data, which showed a lot of layoffs and skyrocketing per hour labor productivity in the very same or immediately succeeding quarters. Those numbers, taken literally, imply that the laid off workers were either producing very little to begin with or they were producing for the more distant future, a’la the Garett Jones hypothesis. The new gdp numbers will imply less of a boom in per hour labor productivity in the period when people are fired in great numbers, though I would be surprised if the final adjustments made this initially stark effect go away. BLS estimates from June 2011 still show quite a strong ZMP effect, although you can argue the final numbers for that series are not yet in. (I don’t see the relevant quarterly adjustments for per hour labor productivity in the new report, which comes from Commerce, not the BLS.) Furthermore there is plenty of evidence that the unemployed face “discrimination” when trying to find a new job. Finally, the strange and indeed relatively new countercyclicality of labor productivity also occurred in the last two recessions and it survived various rounds of data revisions. It would be premature — in the extreme — to conclude we’ve simply had normal labor market behavior in this last recession. That’s unlikely to prove the result.
Most generally, the ZMP hypothesis tries to rationalize an otherwise embarrassing fact for the structural hypothesis, namely high measured per hour labor productivity in recent crunch periods. If somehow that measure were diminished, that helps the structural story, though it would make ZMP less necessary as an auxiliary hypothesis, some would say fudge.
Other parts of the structural story find ready support in the revisions. Real wealth has fallen and so consumers have much less interest in wealth-elastic goods and services. This shows up most visibly in state and local government employment, which has fallen sharply since the beginning of the recession. Rightly or wrongly, consumers/voters view paying for these jobs as a luxury and so their number has been shrinking. Construction employment is another structural issue, and given the negative wealth effect, and the disruption of previously secure plans, there is no reason to expect excess labor demand in many sectors.
In the new report “profits before tax” are revised upward for each year. That further supports the idea of a whammy falling disproportionately on labor and the elimination of some very low product laborers.
Measured real rates of return remain negative, which is very much consistent with a structural story. Multi-factor productivity remains miserably low. In my view, a slow recovery was in the cards all along. Finally, you shouldn’t take any of this to deny the joint significance of AD problems; AS and AD problems have very much compounded each other.