Today’s jobs data release was below our forecast, and that is bad. It is even worse, when one considers the productivity data released earlier in the week. That report showed that productivity has fallen in each of the past three consecutive quarters. This is the most sustained decline since 1979.

Productivity used to have a cyclical component. It fell early in a recession, and it rose early in the recovery. The early-recession fall resulted from falling sales and no employment change. The idea is that businesses see the sales decline, but don’t know if it is temporary. So, they don’t layoff for a while and productivity falls.

The early-recovery productivity growth is similar. A business sees increasing sales, but is unsure if it is permanent. So, they avoid adding to payroll until they are confident that the higher sales will be maintained.

All that went away with the past two recessions. In these recessions, productivity growth was relentless, increasing quarter after quarter. Consequently, our models cannot effectively use the new productivity information. (Don’t ask why. It is a statistical answer.)

Some, very few actually, are discounting the new jobs data, because it included the Verizon strike. We note that it also included the return of Minnesota’s government workers, significantly reducing the Verizon impact.

There are other reasons to be concerned about the new jobs data. A big one is that the previous two months were revised down. June was revised down 26,000 jobs (56 percent) to only 20,000, while July was revised down a whopping 32,000 jobs (27 percent) to 85,000. These revisions imply that the initial estimate is currently biased high, implying in turn that we actually lost jobs in August.

The combination of falling productivity and job losses is a powerful indicator that the second dip may be here or coming very soon.