Paul Krugman offers three good explanations of why today's recessions are involving larger labor losses than in the past. Bubble-based recessions are tougher to get out of, unions are weaker, and many leading firms face more volatility. All taken together, these mean the incentive for labor hoarding is weaker than before (yet Krugman cannot bring himself to mention that labor hoarding models are based on…a zero MP condition. And that a lot of the real work in the account of the cycle is suddenly being done by structural explanations.)
We can all agree with:
1. Some workers are temporarily zero MP because demand is low.
Are there additional factors behind ZMP-like conditions? Those might be:
2. Some workers weren't producing much to begin with and short of retraining they aren't worth so much.
3. Some workers were better than idle, but a one-time, post-firing reorganization of production (e.g., computerizing their task) has since rendered their efforts largely unnecessary. In other words, they are zero MP ex post but not ex ante.
4. Laid-off workers did not start off as zero MP but they will end up as zero MP as their skills and attitudes deteriorate.
5. The Garett Jones hypothesis: many laid-off workers were building up organizational capabilities, and so their perceived MP falls as the discount rates of managers rise.
6. Workers are like advertising: new developments in information technology allow us to better isolate the ones who are not adding value.
The bottom line is that we do not know how long these labor market predicaments will last.
For general background, here is a useful survey of hypotheses for the cyclicality of productivity pre-1990. Here is a good piece on how labor hoarding and productivity measures are related. From Arnold Kling, here are related comments from the structural side.