Statistical Shenanigans At BLS

Non-farm private payrolls for July came in with a surprise 43,000 larger increase than expected. Including new government jobs, the number reported was 75,000 larger than expected.

But, can we really believe the numbers?

This is a Presidential election year, and it turns out that the numbers influence the election results. Good numbers – and a rising stock market – presage a continuation of the party in power, while bad numbers – and a falling market – seem to forecast a turnover to the opposite party.

There seems to be enough motive to fudge the numbers. But, how do you fudge the non-farm payrolls number? People either are working or they’re not working.

One way is through the so-called Birth-Death Model. As the bean counters collect their data, they don’t actually count every single person and business in the country. Instead, they survey a sample – in statistics, a survey of a thousand can come surprisingly close to the results you would get if you counted the whole population.

From the results they get using just the sample, they project what the whole economy is. But the US Economy changes constantly. People leave jobs, and others get jobs. Businesses go out of business, while others start up.

To account for the changing landscape of US businesses, the Bureau of Labor Statistics (BLS) uses its Birth Death Model – they pick a number out of thin air, hoping the number is close – or at least believable. Here are the numbers for the last year or so:

Since April 2015, only three times has the BLS said the Birth-Death Model called for a reduction of the official number, for a total of 280,000 jobs. Against that, there were 13 months of upward adjustments (guesses) for a total of 1,839,000 extra jobs, over and above what their actual survey results said there should be.

If we remove the Birth-Death adjustment for July, instead of beating the estimate by 75,000, it would have fallen short by 37,000. In my book, the numbers are bogus – politically motivated.

After all the hard work of collecting the actual data that the rank and file BLS (and other agency) employees do, their bosses make them look like criminals because of all the ‘phonying’ up that the bosses do.

It likely will not happen ever, but I can dream that someday a new President will have the good sense to investigate and prosecute the perpetrators of fraud at the BLS.