Average Weekly Hours, The Law Of Large Numbers, And An April 618,000 Payroll... Decline?

Tyler Durden's picture

While everyone was focusing on the quantitative component of today's BLS number, it appears what was once again missed in all the noise was the mention of the qualitative aspects of the BLS report: those parts which actually look at the quality of new jobs, not only their earnings power (which as we showed in the breakdown of the April job gains were all toward the lower paying spectrum of available jobs) but also taking productivity and labor demand into the picture. It is here that we find this month's biggest BLS report weakness.

While on one hand, readers are familiar with the following chart showing the constant tapering in growth of wages for production and non-supervisory employees...

... this is primarily driven by the ever-declining wage leverage that US employees have, and wage deflation. Indicatively, in a true recovery, this chart should show a rising line, not a secular decline, and now the second drop in a row, posting it weakest rise since November of 2012.

However the bigger problem with this month's job report was the drop in average weekly hours for all employees (not just those in production and nonsupervisory positions), which posted a surprising and disappointing decline from 34.4 to 34.6 on expectations of an unchanged number. This amounts to a 12 minute shorter workweek on average for the entire US labor force.

Hardly notable?

It is when one considers that there were 135,474,000 full time Establishment Survey employees in April (rising by the much trumpeted 165,000), all of which worked on average 34.4 hours (down from 34.6 in March) according to the BLS. Multiply these together and one gets 4,660,305,600 total hours worked weekly in April, a drop of 21,385,800 million hours from the 4,681,691,400 total hours worked each week in March.

Then apply the average hourly wages of $23.83 in March and $23.87 in April, and the total weekly wages paid out in March ($111.565 billion) compared to April ($111.231 billion) amounted to a drop of $323.2 million on a weekly basis.

Had the average weekly hours stayed flat as expected, this number should have been an increase of $323.5 million or a $646.8 million swing!

In other words, the US economy added 165,000 jobs and yet US businesses paid $323.2 million less in total wage compensation: only the second time there was a decline in the gross total monthly wages paid in 2013.

What does this mean for the bottom line?

Well, had the BLS reported flat average weekly hours worked at 34.6 as Wall Street had expected, while companies were paying out the same amount of hourly wages in April, the result would have been that instead of the BLS reporting a 165,000 increase in jobs, it would have had to report a drop of, drumroll, 618 thousand workers, or total April workers of 134,690,913: a 783 thousand negative worker swing, more than wiping out not only all the gains of April, but all prior upward monthly revisions as well.

MarketWatch adds some granularity:

Some analysts applauded the 29,000 gain in retail-sector jobs in April as a sign that consumer spending is holding up well in the face of the fiscal drag caused by the tax hikes and government spending cuts.


But aggregate weekly hours worked in retail plunged by 0.7% in April, which is the equivalent of cutting 11,000 jobs. Suddenly, the report doesn’t look so rosy.

So productivity increases? Economic Recovery? Or just the BLS (such an expert in dropping the unemployment rate at the expense of the Labor Force Participation rate) showing its exquisite familiarity with the law of large numbers?

You decide.

(P.S. this entire analysis is trivial as the entire difference "on the margin" in the jobs number is based on seasonal adjustments which have a +/- 200,000 error rate. However, the algo response to today's BLS flashing red headline - all that really matters - would have been vastly different had the BLS reported a negative number, which just also happens to fall in the margin of error).