By Matt Gibson/Shutterstock.com
This past friday, December 6, the BLS reported a jobs gain in November of 266,000, surprising just about every estimated forecast. At least 70,000 or so were not actually ‘new’ jobs, but the return to work of GM auto workers and related auto industry suppliers. But the 190-200,000 net jobs gain reported was nonetheless way above any expectations.
What’s the contrary evidence?
The independent enterprise payroll company, ADP, which reports out job numbers a few days before the BLS every month, based on its tens of thousands of companies it gathers data from, had estimated a mere 67,000 jobs net gain. 266,000 v. 67,000 was perhaps the biggest gap ever between the two reports–one private and the other government. Thus far no explanation has been provided for the massive gap in job forecast, by either ADP or the government.
Also worth noting is that the BLS, the government, this past summer adjusted its total jobs number for 2018 by reducing the total job creation in 2018 by a massive 500,000. How reliable are the monthly numbers when they have to be adjusted by 40-50,000 a month on average? Recently the government announced it expected that 2019 total jobs would likely have to be reduced by at least 500,000 when it did its annual adjustment and reductions again this coming February 2020. Again, that makes the monthly numbers suspect. Why the massive adjustments the past two years? Are the monthly numbers not reliable for some reason? What’s the reason?
It’s important that readers understand that the monthly jobs numbers don’t represent actual jobs. They are a statistic. That means the raw actual number of jobs are not what’s reported. It’s a statistical manipulation–based on a series of complex assumptions and even more complex mathematical formula adjustments of the raw jobs data–that gets reported monthly as the job numbers.
Moreover, the monthly numbers reflect jobs, not actual workers getting new employment. There may be workers adding second and third jobs, reflecting the jobs increase, but not actual new employment increase. The US Labor Dept. claims it captures added jobs by those already employed, but the numbers suggest its methodology may not be that accurate. The US labor markets have radically changed since the late 1990s and the BLS methods may no longer be accurate for picking up 2nd and 3rd jobs. The changes also suggest that maybe the government’s statistical adjustments for seasonality are not that accurate any more.
There has been greatly volatility in the monthly jobs numbers this past year. Some months well below 100,000, representing the fact that more new workers are entering the job market and not finding jobs. It takes at least 125,000 to 150,000 to absorb all new entrants. The great volatility suggests the BLS is not picking up the seasonal changes month to month very well; and only does so with its annual adjustments to the prior year’s data–hence the 500,000 annual reduction in the total numbers of jobs for the year.
Then there’s the adjustments in the statistical manipulation of the raw jobs data for what’s called ‘New Business Formation’. This is the adding of jobs to the raw data numbers by assumptions of new business formation. It’s assumed every month that several hundred thousand net new businesses are formed. Each has a number of new jobs assumed associated with the formation. Problem is that the New Business Formation is from 6-9 months previous to the current month. In other words, the BLS assumes a net gain of jobs from March-May 2019, adds that to the raw jobs data for November, then does a number of statistical manipulations on the combined actual jobs for November plus 9 month lagged jobs from New Business Formation, and that’s what is reported out as the November jobs number. Except in times of deep recession the added jobs from business formation are always positive. So the raw jobs data is always increased for the month–even before other statistical manipulations of the combined raw data are performed for that month, i.e. for seasonality and other reasons.
It’s difficult to assume that’s all OK with the monthly jobs numbers reporting for these and related additional reasons.
Jobs and employment lagged the general direction of the economy. But that real economy has been contracting in various dimensions. For example, net new business investment has been contracting ever since the 2nd quarter of 2019. Manufacturing has been contracting for four consecutive months. Commercial construction for longer. Residential housing construction contracted for most of 2018-19, rose slightly for several months due to interest rate cuts, but is contracting once again. Export driven job creation is unchanged, as net exports (exports minus imports) has hovered steadily around a negative $50b a month throughout the period of Trump’s trade war for the past 20 months. So where’s the real economic growth coming from that would produce 200,000 plus jobs? If employment is a lagging indicator, it’s certainly lagging as never before.
We’re told that it’s the service sector and household consumption that’s holding up the real economy. But services growth has been slowing its growth rate in recent months as well. Consumption is being driven by the wealthiest 10% of the labor force (where wage gains for professionals, tech, healthcare, managers, etc. are concentrated) and by record credit creation for the rest of consumers. But that hardly explains the unexpected big surge of 266,000 jobs last month. Neither services nor consumption explains companies hiring by that big an increase in one month. Some argue that its the government growth in census workers, but reportedly that was largely concluded by October.
So where’s the explanation for the big gap in BLS November jobs numbers from the private payroll estimates of ADP? The ADP numbers are actual payroll data. The BLS government numbers are the result of statistical operations and manipulations piled on one another to get a ‘statistic’ and thus estimate of actual jobs.
This writer still argues that the labor market changes in the US have caused the BLS government methods and statistical manipulations to become less accurate and more volatile from month to month (only adjusted more accurately when the annual reductions are made by half a million jobs or m ore). The BLS seasonality and New Business Formation assumptions are less and less accurate. And its methods for estimating 2nd and 3rd jobs taken on by the already employed over-estimate the job numbers monthly, and the government still grossly underestimates the effect of millions it dumps into its ‘catch-all’ category of the ‘missing labor force’.
For the real jobs numbers we’ll have to wait until February 2020 when the BLS does its annual adjustment for 2019–and likely reduces again the job number reported for November and 2019 by another 500,000 or more!
Dr. Jack Rasmus is author of the just published book, ‘The Scourge of Neoliberalism: US Economic Policy from Reagan to Trump’, Clarity Press, January 2020. The book is available at 20% discount from his blog, jackrasmus.com, and website, http://kyklosproductions.com.