As hurricanes pass, their effects are reflected in economic data, though not in the way one would expect. Last Friday’s payroll report showed a contraction of -33,000 jobs, and the Bureau of Labor Statistics rolled back recent prior reports by another -38,000. Those additional revisions were prior to the hurricanes.

The household survey, the data series that provided the unemployment rate data, was spectacularly strange. It showed an increase of +906,000 employed workers, which sent the unemployment rate down to 4.2%. For the first time since May 2014, the labor participation rate hit 63.1% (it’s still three percentage points below historical norms). 

That +906,000 is a huge increase, and can’t really be believed. We’ve had changes to the number of employed before of similar magnitude and they have been related to specific events such as the ends of strikes and weather-caused disruptions. This increase is despite the decreases in the payroll report for restaurant and similar workers because they were told not to report to work. It takes time for these distortions to work themselves out in the data just in a definitional sense. The same person can be considered unemployed in the payroll survey and be classified as employed in the household survey. In situations like these, the BLS issues a special statement from its commissioner to reassure everyone that they have procedures for these situations (they do, and they are good and tested), but the statement did not really illuminate or clarify much of anything. The Wall Street analysts are always looking forward to the next report of data, but this is one of the few times where it’s really necessary.