What’s the best measure of unemployment during a pandemic? All of them

On the heels of last week’s surprisingly positive May jobs report, the controversy over the accuracy of the official unemployment rate has showcased the difficulty in assessing how the pandemic has affected the labor market. The standard unemployment metrics are fairly accurate in normal situations, but they haven’t been as useful due to the extreme employment shifts we’ve seen over the last three months.

Thankfully, researchers are developing additional metrics that can help us understand the situation better. Here is what a few of them reveal.

The official unemployment rate, U-3, is one of six developed by the Bureau of Labor Statistics (BLS) during its last major labor survey redesign in 1994. U-3 focuses on counting the jobless workers who are most likely to contribute to the supply of labor. For this reason, a person is only counted as “unemployed” if they are currently jobless, have actively sought work in the last four weeks and are available to start working.

Obviously, this definition leaves some workers out. The “marginally attached” unemployment rate (U-5) counts those who are available for work and have sought it within the last year, but who haven’t searched in the last four weeks. Meanwhile, the “underemployment” rate (U-6) adds those who are working part-time (for economic reasons) but want full-time employment.

But some people still slip through the cracks. Along with Joint Economic Committee economist Scott Winship, I’ve previously suggested an expansion to U-5 called the “comprehensive jobless rate.” It counts everyone who currently wants a job, regardless of when they last looked for work or their availability. It rose from 6.3 percent in February to 18.9 percent in May.

But the confusion caused by COVID-19 means that even these alternative measures of unemployment aren’t enough. Millions of workers have been misclassified as “employed, but absent from work” rather than “unemployed, on temporary furlough.” The BLS has acknowledged the error, but its policy of not changing survey respondents’ answers – to avoid the appearance of politically-motivated changes – means that nearly all BLS unemployment estimates are missing a substantial number of workers. 

If the 4.9 million misclassified workers were added to the ranks of the unemployed, the U-3 unemployment rate seen in most headlines would be around 16.4 percent rather than 13.3 percent.

Millions more workers haven’t been counted because they don’t meet the conditions to be considered participants in the labor force. For example, some who have lost jobs aren’t immediately seeking re-employment. During the data collection week for the May jobs report, over 30 million workers filed for unemployment insurance benefits — 9 million more than the labor surveys counted as officially unemployed.

Government-mandated business closures and stay-at-home orders may be partially to blame for people not looking for work, along with the federal bonus to unemployment insurance and the relaxation of the job-seeking requirement typically required for unemployment insurance eligibility.

Regardless, these workers are likely to “rejoin” the labor force soon and so should be counted as unemployed. If the 6.3 million who have “left” the labor force since February were counted and combined with the misclassified workers, the official unemployment rate would be around 19.5 percent. 

I call this metric the “pre-pandemic comparable unemployment rate” (U-3_PPC) because it provides a current unemployment rate that we can compare with February’s 3.5 percent unemployment rate.

Likewise, Chicago Federal Reserve economists Jason Faberman and Aastha Rajan have suggested a measure of slack labor: “U-Cov.” It adds workers on unpaid leave and people who say they want a job but aren’t counted as unemployed (similar to the comprehensive jobless rate) to the underemployment rate (U-6). U-Cov started out at 10.4 percent in February and, after a spike in April, decreased slightly to 27.6 percent in May. 

Former Treasury Department economist Ernie Tedeschi has proposed a more holistic labor underutilization rate, “NPOP.” It combines (1) unemployed workers, (2) nonparticipants in the labor force, and (3) all under-employed workers (those working part-time who’d prefer full-time), providing an upper-bound estimate of pandemic-related labor underutilization.

I estimate May’s NPOP rate reached 56.8 percent, but this shouldn’t be a cause for alarm. Because NPOP counts students, retirees, and stay-at-home parents – essentially any person not working full-time or voluntarily part-time – it will naturally be much higher than other unemployment metrics. Rather, it’s the change in NPOP that helps us understand the labor market. It suggests that the pandemic has reduced the labor market output of 7.8 percent of the working-age population since February.

Each of these unofficial unemployment metrics is intended to complement, not replace, the existing measures. They provide new perspectives to help us understand the bigger picture a little bit better. Right now, we need all the help we can get.

Michael D. Farren is a research fellow with the Mercatus Center at George Mason University.