U.S. Government Acts to Reduce Distortion in Jobless-Claims Data
(Bloomberg) -- The U.S. Labor Department said Thursday it’s changing how it adjusts weekly jobless-claims figures for seasonal swings after the pandemic’s surge in unemployment led to an unusually large divergence between adjusted and unadjusted figures.
Unadjusted jobless-claims figures experience major fluctuations around holidays such as Christmas and other annual events like summer auto-factory shutdowns, when companies add and lay off large numbers of workers. The Labor Department uses a process called seasonal adjustment to smooth the figures for these anticipated changes.
While there are typically occasional challenges with the process, the unadjusted and seasonally adjusted jobless-claims figures tend to move in tandem.
But the pandemic has wreaked havoc on the Labor Department’s typical method of seasonal adjustment, in some cases severely overshooting or undershooting the non-seasonally adjusted figure. At one point in March, the adjusted claims figure was more than 851,000 higher than the unadjusted figure. In the last two months, the series have occasionally moved in different directions on a given week.
Many economists have increasingly focused on unadjusted figures in recent months.
In statistical terminology, the Labor Department has traditionally used “multiplicative factors” to adjust the claims data. That means multiplying the unadjusted number by a certain percentage each week. The alternative is an “additive factor,” which entails adding or subtracting a number from the claims figure to adjust it.
When the economy is relatively stable, the multiplicative factor is “generally preferred over the additive option,” the Labor Department said Thursday. But when there’s a sudden, large shift in a series -- like the pandemic driving initial claims well into the millions, from just over 200,000 -- a multiplicative factor can lead to systematic over- or under-adjustment.
Starting in next week’s release, the Labor Department will use additive factors, since in cases such as now, “they tend to more accurately track seasonal fluctuations in the series and have smaller revisions,” the department said in Thursday’s jobless claims report.
“Seasonals are just not designed for this kind of a crisis, so the additive makes more sense in this circumstance,” said Erica Groshen, a former Bureau of Labor Statistics commissioner. “You can see that in the extraordinary circumstances where we are now that a percentage would probably mislead you.”
Whether already published claims data that used the multiplicative factor will be revised is unclear, but that likely won’t happen until the beginning of 2021 when prior years of seasonally adjusted estimates are subject to revision.
©2020 Bloomberg L.P.