The Bureau of Labor Statistics (BLS) surprised the markets and most economists Friday with an announcement that the unemployment rate fell from 14.7 percent in April to 13.3 percent in May. President Donald Trump had the temerity to boast that George Floyd, the unarmed black man killed by a white police officer in Minneapolis, was “looking down from heaven” to admire those numbers. “This is a great day for him,” Trump said in the Rose Garden. “This is a great day for everybody.”
Beyond the president’s remarkable and offensive invocation of Floyd, there is good reason to be skeptical of his administration’s latest report.
BLS asserts that 20,935,000 Americans were unemployed in May, attributing the decline to employers adding 2,509,000 jobs in May as states allowed “a limited resumption of economic activity.” But it’s not that simple. The unexpected decline in the jobless rate is based on a survey of businesses and households conducted over the week of May 16th, and BLS has also reported that 29,965,415 Americans received unemployment insurance benefits in the same week. That’s a gap of 8,980,415 people, enough to raise the May jobless rate 5.7 percentage points to 19.0 percent and much closer to economists’ expectations.
How did the Labor Department come up with such a low figure? Part of it is the Payroll Protection Program: The BLS counted anyone who employers say were still being paid as employed “even if they were not actually at their jobs.” Part of it involves how the BLS treats the many millions of people now on furlough and not being paid. They are considered “unemployed on temporary layoff.” But if BLS expects them to return to their old jobs, based on the survey, they do not count among the unemployed.
Those judgments are left to the BLS’s discretion, and the agency does not issue any data on those decisions. Yet a recent analysis from the Becker Friedman Institute at the University of Chicago estimated that 42 percent of people furloughed by the COVID-19 crisis will never get their old jobs back, and only 30 percent of those laid off will land new jobs later this year. With 30 million people receiving unemployment benefits, that analysis is more consistent with unemployment at 19.0 percent today than 13.3 percent. BLS itself acknowledges in its May report that counting those on furlough as unemployed would raise its official rate to 16.3 percent.
Finally, part of the 9 million-person discrepancy may involve how BLS decides whether a person who is not working is still part of the labor force. If you are out of work but didn’t look for a job over the previous four weeks, or couldn’t work for family reasons such as caring for children who are no longer in school, BLS says you are out of the labor force. If you’re not in the labor force, you’re not unemployed.
Again, it is BLS’s judgment call—and BLS clearly has decided that millions of people who lost their jobs because of COVID-19 shutdowns do not count as unemployed. In May 2019, BLS reported that 96,207,000 working-age Americans were not in the labor force; today, BLS reported that a year later, that number had jumped by 5,865,000 to 102,072,000. For reference, the BLS found that 95,636,000 working-age Americans were not in the labor force in 2019, down 80,000 from 2018. Those 5,865,000 people could explain nearly two-thirds of the 9 million-person gap between those receiving unemployment benefits and those counted as unemployed.
Sometimes, unexpected events produce wildly inaccurate results from computer models that assume the unexpected won’t happen. Perhaps there is a glitch in the BLS unemployment model that couldn’t accommodate disruptions from the COVID black swan. We also have seen the Trump administration interfere with scientific analysis in other areas, such as when it directed what models the U.S. Geological Survey could use for the mandated National Climate Assessment. It would be unprecedented for the White House to pressure the Labor Department to produce a jobless report that serves Donald Trump’s needs and desires. But unlike the May unemployment report, it would not surprise anyone.