Fewer Hands, Longer Hours

The work file points in both directions at once. Hourly pay stands at 37.62, up 1.15 from 36.47 a year ago; hours worked hold at 34.3, the top of the two-year window at z of 0.94. Building jobs sit at 8343, up 82 from 8261, and temp jobs at 2505, up 10.8 from 2494.2. By the plain account, work pays more for more time.

The same file shows employment at 158,858, down 23 from 158,881; factory jobs at 12,611, down 14 from 12,625; public jobs at 23,270, down 315 from 23,585. Fewer people are working, and each of them works longer for more. That is not strength or weakness. It is a split between the price of an hour and the count of hands, and today's data cannot say whether it is composition, a change in who holds the jobs, or condition, a change in what an hour is worth.

Don takes pay and hours as confirmation and sizes up; the slip to 158,858 from 158,881 must be noise. His position never names what would falsify it, but today the file does that for him. Franc sees a relationship altered, and his version requires him to name which one. The candidate is hours against headcount: before, they moved together; now hours top their window while the count stalls.

He usually declines to supply the before and after, and without it the position is a mood, not a claim. The monthly payrolls release, on its published schedule, prints pay and headcount together. If the two series resolve in the same direction, the split was composition: Don concedes nothing if it resolves upward, and everything, the flat count having been signal, if it resolves down.

If they diverge again, condition is the working answer, and Franc owes the cheaper concession already past due, the name of the relationship he says has changed.