Don’t believe the overstated ‘headline estimates’ of unemployment 

Since COVID-19 deaths began to rise in mid-March, people have awakened every Thursday to the latest report of the number of unemployed associated with virus-related shutdowns. A recent “headline estimate” of job loss was 40.8 million people, but a more reasonable and accurate number is less than half that. The overstated loss in employment of 23 million people is large and grows every week.

The headline estimate is based on adding the number of “initial claims” for unemployment compensation, filed with state unemployment insurance offices each week, to the total from the week before. The report is issued by the U.S. Department of Labor’s Employment and Training Administration for the week ending the previous Saturday. But each week, the same report includes the continued claims for unemployment — called insured unemployment — which includes those who claim they were unemployed for at least a week and who are due unemployment compensation.  

People who file initial claims are often subsequently revised off the rolls because their claim was invalid or because they obtained jobs before they could claim their first weekly benefit. Each week some of the past unemployed drop out of the “continued claims” data because they find new jobs or, these days, are called back to existing ones, and are no longer eligible for unemployment compensation. 

Another problem with the headline estimate of lost employment is that it uses seasonally adjusted data, which do not count actual people, but rather estimates that add or subtract people based on normal seasonal patterns. For the headline measure reported over the past 10 weeks, the sum of initial claims has exceeded the actual number by up to 3.2 million people each week. Applying a seasonal adjustment to data largely based on an unprecedented government shutdown of businesses yields a meaningless result. Unadjusted data are used here.  

Continued claims data, released with a one-week delay, show a much smaller number of the growing unemployed. According to that measure, the number unemployed since March 14 — when it equaled 2.1 million — rose to 19.1million in the week ending May 16, a rise of 17.0 million. This is less than half the headline number, but it is for a period ending a week earlier. For the same week, May 16, the headline sum of initial claims was reported to be 38.6 million, about 21.6 million more than the estimated 17 million based on the rise in continued claims. 

Some part of initial claims for the week of May 16 will be added to continued claims for the week ending May 23. Based on the past relationship between initial claims and continued claims, the continued claims for the latest week of May 23, are estimated by the author to be 17.6 million, which will be only 43 percent of the widely reported headline estimate of 40.8 million people.

So, did unemployment rise by 40.8 million because of the surge in coronavirus over the past 10 weeks? Absolutely not. Because of factors such as seasonal adjustment, mistakes by initial claimants, errors in whether workers were truly unemployed, or workers obtaining new jobs or getting their former jobs back, continued claims data suggest the correct rise in unemployment as of May 23 was only about 17.6 million. Indeed, a recent Federal Reserve survey indicates that 9 out of 10 of those who lost their jobs because of the coronavirus shutdown expect to be called back to their old jobs. The slower the reopening of their employer’s businesses, however, the less likely are those expectations to be realized.

Nonetheless, the rise in unemployment is still massive — though the evidence here provides a strong warning that fears of Great Depression levels of unemployment are not likely to occur because of the pandemic. Instead, the unemployment rate for April rose to 14.7 percent from March, still well above the post-World War II peak of 10.8 percent at the end of 1982, instead of an excess of 20 percent expected by many observers. It is not likely to rise much further from here, if at all. 

Moreover, in 1983, the unemployment rate fell in the 12 months after its peak, largely because of the Reagan tax cuts kicking in, by 2.5 percentage points. This time, the relaxed constraints of the government-induced rise in the unemployment rate should bring it down much more rapidly.

The old Chinese curse, “May you live in interesting times,” is wholly inadequate to describe the surge in deaths from COVID-19 to more than 100,000 in the U.S. in 10 weeks (from March 17, when 121 had died, to 100,572 on May 26, according to Worldometers.info), the near shutdown of the economy, with a recession decline at a 5 percent rate in the first quarter the year, a rise in unemployment to 4.4 percent in March and even larger rise in April to 14.7 percent. We need not overestimate the pain, at least in terms of the size of the rise in the number unemployed. 

John A. Tatom is a fellow in the Institute for Applied Economics, Global Health and the Study of Business Enterprise, Johns Hopkins University, and a former executive director and head of Country Risk and Limit Control at UBS AG in Zurich.