Why US Job Reports Make No Sense

Why US Job Reports Make No Sense

The following is based on the extraordinary work of Jeffrey Snider of Alhambra Investment Partners. The unabbreviated version is found here: http://www.alhambrapartners.com/2015/05/06/the-most-intrusive-thorn-in-the-side-of-anyone-proclaiming-qe-worked-and-the-economy-has-taken-off/

For more than a year, the US Bureau of Labor Statistics (BLS) has been saying that we are in the middle of the best jobs market in decades. Never mind that the percentage of working age people with jobs has been falling month after month while the Establishment Survey showed cumulative job gains by the millions. Never mind that, as the jobs market was purportedly streaking ahead, incomes and spending inexplicably were not. The Fed saw what it wanted to see in the Jobs data and declared that slack in the labor market was lessening...proof positive of economic recovery.

As Snider noted yesterday, the Establishment Survey does not actually measure payrolls. It subjects a sample to statistical manipulations based on an assumed benchmark that is set mostly by trend-cycle estimates which, in turn, probably rely heavily on jobless claims. Meanwhile, the Bureau of Economic Analysis (a different agency) measures economic output. These two data series should obviously relate...more jobs should mean more output, unless businesses managers have suddenly lost their minds. But they do NOT relate.

Productivity is the measure that should relate the BLS jobs data to the BEA output data. The BLS takes a pared back version of the BEA's GDP, separates out "private output" and subtracts its estimate of total hours worked. The remainder is productivity.

Here is the hours worked side of the equation starting in 2009, which supports the view that QE has been highly effective in promoting labor utilization.

But if we look further back, it is clear that hours worked have not recovered the trend; they are barely back to the 2000 peak.

Meanwhile, total output per hour worked is in a downtrend.

As Snider notes, the mainstream interpretation is that the Establishment Survey is correct: businesses are hiring but not getting much out of doing so. Hiring is proceeding as productivity declines? That just doesn't make sense.

Three out of the past five quarters have reported deeply negative productivity. Assuming that hiring is robust and output is not, the government statisticians are forced to conclude that productivity is cratering and private sector mangers have lost their minds. That's the only way to balance the numbers.

As Snider notes, the problem is really sharpened in Q4 2014 where we are supposed to believe that businesses were hiring at exceptional rates, leading to a nearly 5% increase in labor hours, but output grew only slightly more than half of that. The resulting productivity calculation was -2.2%. For Q1, output was slightly negative but business still expanded labor hours?

The discrepancy in the numbers actually goes back five quarters.

The BLS measure of employment does not sync with the BEA's weaker measure of output. Rather than revise one or the other, productivity is made negative (as the remainder) to preserve both figures. Output cannot be revised upward because calculated inflation in 2014 and 2015 is almost zero, suggesting, as Snider says, that the labor calculations are far too robust.

What happens if the productivity rate is held at its average for 2004-07 and output is unchanged? Hiring over the last five quarters is reduced by 75%!

Productivity historically has been much better than in the 2004-7 period. Using the average productivity from 1992 to 2000 would yield approximately no growth in labor over the past five quarters:

Something is very wrong in the way job gains are measured in the US economy. It isn't just the definition of who is employed and who is not, important as these definitions are, nor is it simply a failure to recognize the relevance of declining labor market participation rates and the growing proportion of part-time workers. It seems that the BLS is counting workers who are not there.

Something is very wrong in the way job gains are measured in the US economy. It isn't just the definition of who is employed and who is not, important as these definitions are, nor is it simply a failure to recognize the relevance of declining labor market participation rates and the growing proportion of part-time workers. It seems that the BLS is counting workers who are not there.

‍