A Full Week on a Thinner Roster

Hours worked stands at 34.3, the top of its rolling window, which spans 34.1 to 34.3, a z-score of 0.94 against a mean of 34.25. That looks like a labor market refusing to bend. It is not obviously that. The same tape shows Employment at 158,858, down 23 from 158,881, Factory jobs at 12,611, down 14 from 12,625, and Public jobs at 23,270, down 315 from 23,585.

Fewer people are on payrolls, and each of them is working a week at the long edge of recent habit. Two facts, one economy, and the question is which one is telling the truth about demand. Don takes the hours figure as confirmation: employers short on labor stretch the week before they hire, so the headcount slippage is noise and the edge is the signal.

That position spends the 315 and the 23 and the 14; it requires all three to be statistical weather rather than a trend. Franc takes the divergence itself as the fact: before, hours and headcount rose and fell together, and now they point apart, which means the relationship between the two has changed. That position spends a name for the new relationship.

Without one, it is a description dressed as an explanation, and the divergence could still be two series catching up to each other. What the data cannot settle is what holds the week at 34.3 while payrolls shrink. The monthly payrolls release is the next test. If hours return inside the window, Don concedes the edge was noise and Franc concedes the marriage held.

If the week stays at the top while headcount falls again, Don's stretch-before-hiring account gets one more month, and Franc must finally say which relationship broke and when.