The stagflationary report on first-quarter gross domestic product from the Bureau of Economic Analysis got plenty of attention, but the Bureau of Labor Statistics recently dropped a bigger bombshell, and no one seemed to notice.

It turns out the economy may be losing jobs, not adding them.

The BLS provides monthly estimates for the number of nonfarm payrolls across the country through a survey of over 600,000 businesses. According to that data, payrolls increased by 640,000 in the third quarter of last year.

But now, the BLS has published its Business Employment Dynamics, or BED, report from that same period, and it shows a drop in private payrolls of 192,000. That’s a mammoth difference of 832,000 for the months of July, August and September.

This revelation raises two questions: What accounts for this disparity, and which metric is to be believed?

First, the two reports count slightly different things. The monthly jobs estimates published by the BLS are the number of nonfarm payrolls, which excludes the agricultural sector but includes government workers.

Conversely, the quarterly BED report counts payrolls at 9.1 million private businesses, so it excludes government workers.

Despite their differences, the two reports have a tremendous amount of overlap in what they are trying to estimate, which is why they typically trend closely together over time. The recent disparity between them is highly unusual.

While government payrolls increased greatly in the third quarter of last year, jumping by 146,000 in just those three months, that accounts for less than one-fifth of the difference between the two reports from the BLS.

It is worth noting that the BED report comes from the quarterly census of employment and wages, which covers more than 95% of jobs in the U.S. and includes about 12 million businesses. That is over 17 times the sample size of the dataset used to compile the monthly estimate of nonfarm payrolls.

Not only does the BED have the law of large numbers on its side, but the monthly job estimates last year also contained two red flags.

First, they were nearly all revised down. While revisions are standard operating procedure, their direction tends to be random. Repeated downward revisions, however, are more consistent with economic downturns, as happened in the Great Recession.

Second, there has been an unprecedented divergence between the survey of businesses and the survey of households, the two datasets used to compile the monthly employment situation published by BLS. The household survey shows net job losses since last August, in stark contrast to the 1.7 million-job gain shown in the establishment survey.

These facts indicate the BED report is likely a better indicator of how many jobs the economy is adding — or losing, in this case.

In addition, the labor market weakness being sounded by the BED report should have started ringing alarm bells in January. That was when the second-quarter data became available, and it showed an increase in private payrolls of about half what the monthly job reports showed, even after those monthly figures were already heavily revised downward from their initial readings.

Many pundits have puzzled over why polling data shows such negative views about the economy when so many jobs are being created. If the latter is untrue and jobs are being lost, that certainly helps explain the former.

Originally Published by The Washington Times