The Bureau of Labor Statistics has revised down its estimate of total U.S. job creation by 501,000 over the period from March 2018 to March 2019. The change results from BLS’s annual benchmarking process, in which BLS recalculates previously-reported monthly employment estimates by checking its statistical samples and modeling against administrative data from state unemployment systems.
BLS’s monthly job creation/loss data is based on a survey of 142,000 employers — so called “establishments” — with 689,000 worksites, that provide payroll data to BLS. From that representative sample, BLS extrapolates to estimate the total number of jobs at the country’s 9 to 10 million employers.
John Stewart, BLS’s supervisory economist for benchmarking the Current Employment Statistics survey (CES), said “to correct for any errors that have accumulated in our survey estimates over the course of a year,” agency statisticians compare the monthly statistical sampling against an actual count of employees that comes from administrative records of state unemployment insurance systems, which capture more than 98% of all U.S. employees.
This is the preliminary revision to the past year’s job numbers; a final revision will be released in February 2020. Stewart said the preliminary revision is typically close to the final revision, varying by an average of 41,000 over the past ten years.
This year’s preliminary revision of 501,000 jobs is larger than any since 2009, when the total number of jobs in the economy was revised down by 902,000. The revision for March 2017 to March 2018 was a decline of 16,000 jobs. (See Table 22 here for annual final BLS revisions, 1979-2018.)
Economist Martha Gimbel at Indeed.com said that revisions are typically smaller when the economy is stable or growing modestly, as it has been in recent years. “But they [revisions] can be bigger when the economy is at a turning point,” she said. “It’s easy for the surveys that BLS does to miss something.”
As an example, she pointed to the tumult in retail, with store chains closing and layoffs mounting. Retail jobs were adjusted down by approximately 180,000 in the latest revision.
Heidi Shierholz of the Economic Policy Institute served as chief economist at the Labor Department from 2014 to 2017. She said one source of error in BLS’s job-number estimates comes from the way BLS statisticians estimate the number of businesses that are created and how many close.
“Typically when firms go out of business, BLS doesn’t actually know that they’re out of business, they just think they’re not responding to the survey. So they can’t really know for sure if a business has died.”
UC Berkeley economist Jesse Rothstein said this preliminary downward revision of March 2018 to March 2019 job growth is sobering, but not alarming:
“Maybe it gives us a little bit more concern that we might be cresting and starting to head down into a recession soon. But I don’t think this changes that likelihood all that much.”