While it's not the first time we have observed the BLS manipulate data (the last time was in "This Is What Happens When The Bureau Of Labor Statistics Is Caught In A Lie"), never before had we actually caught the Bureau Of Labor Statistics openly fabricating data. Until now.

As reported earlier today, in one of the most closely watched statistics in today's payrolls report, the BLS reported that the annual increase in Average Weekly Earnings was a whopping 2.9%, above the 2.5% expected, and above the 2.5% reported last month. On the surface this was a great number, as the 2.9% annual increase - whether distorted by hurricanes or not - was the highest since the financial crisis.

However, a problem emerges when one looks just one month prior, at the revised August data. 

What one sees here, as Andrew Zatlin of South Bay Research first noted, is that while the Total Private Average Weekly Earnings line posted another solid increase of 0.2% month over month, an upward revision from the previous month's 0.1%, when one looks at the components, it becomes clear that the BLS fabricated the numbers, and may simply hard-coded its spreadsheet with the intention of goal-seeking a specific number.

Presenting Exhibit 1: Table B-3 in today's jobs report. What it shows is that whereas there was a sequential decline in the Average Weekly Earnings for Goods Producing and Private Service-producing industries which are the only two sub-components of the Total Private Line (and are circled in red on the table below) of -0.8% and -0.1% respectively, the BLS also reported that somehow, the total of these two declines was a 0.2% increase.

Another way of showing the July to August data:

• Goods-Producing Weekly Earnings declined -0.8% from $1,118.68 to $1,109.92
• Private Service-Providing Weekly Earnings declined -0.1% from $868.80 to $868.18
• And yet, Total Private Hourly Earnings rose 0.2% from $907.82 to %909.19

What the above shows is, in a word, impossible: one can not have the two subcomponents of a sum-total decline, while the total increases. The math does not work.

This, as Zatlin notes, undermines not only the labor inflation narrative, but it puts into question the rest of the overall labor data, and whether there are other politically-motivated, goalseeked"spreadsheet" errors.

We have sent an email to the BLS seeking an explanation for the above data fabrication, meanwhile here is what likely happened: a big, juicy fat-finger error, whether on purpose or otherwise because if one looks at the finalized July weekly earnings of $907.82, it's precisely the same as what the August preliminary wage number was as released last month, also $907.82. For the excel fans out there, it means that the August totals were simply hard coded when the BLS shifted cells in the spreadsheet, becoming July.

Of course, if the BLS confirms that this was a transposition fat finger error, it would also imply that the August number is in fact, the September data, a rather massive mistake which today has had a impact on trillions dollars worth of assets.

Source: BLS