The latest labor market data show that aggregate hours worked fell again in August. This means that LUCY, my indicator of labor capacity utilization, also dropped.

The longer that this productivity-cushioned recession continues, the more of a puzzle it presents to macroeconomists. The issue is this: if productivity is rising much faster than real wages, why aren’t firms hiring more workers?

Let me sketch a couple of possible answers briefly.

The one-sector model

If we think of the economy as consisting of a single labor market, then I believe that the only way to explain recent economic behavior is to stress the distinction between average productivity and marginal productivity. That is, one can argue that even though average productivity is rising, marginal productivity is falling.

Consider a typical industry with new-economy characteristics, which means that a lot of labor goes for research, trial-and-error, marketing, and other expenses that do not contribute to output in the short run. The perceived marginal product of these activities may have fallen, because corporations have changed priorities.

With the stock market down, firms may place a higher value on retained earnings, which would lead them to reduce overhead. Another factor may be Sarbanes-Oxley, the bill that made corporate boards and executives more accountable for certifying earnings. I have been accumulating some anecdotal evidence that this is having a real dampening effect on the risk-taking of companies. Workers that are involved in experiments to build a better mousetrap, or to come up with marketing strategies for such a mousetrap, are now seen as expendable.

The two-sector model

The two-sector model is one in which in one sector of the economy has productivity growing faster than demand, while in another sector demand is growing only as fast as productivity. The first sector is shedding workers, while the second sector is standing pat. On average, productivity goes up and labor demand goes down.

For Discussion. What empirical evidence would you look for to examine the validity of either of these models?