The shutdown of the American economy due to the COVID-19 pandemic has focused considerable attention on data reported by the Bureau of Labor Statistics (BLS) and the Employment and Training Administration (ETA) of the US Department of Labor. Reports by these agencies over the past few months have indicated unprecedented short-run changes in employment and unemployment, but at times these reports appear confusing and conflicting. An example is the substantial disparity between the number of unemployment insurance (UI) benefit recipients and the number of unemployed workers. The latest Employment Situation Report from the BLS indicated there were 20.51 million unemployed workers during the week of May 11-15. During the same week the ETA reported that 30.17 million workers were collecting UI benefits. For the first time in history there are many more workers collecting UI benefits than there are unemployed workers.
This blog post explains the apparently conflicting unemployment figures. Because of Federal changes in the UI system enacted during the pandemic: (i) underemployed workers have much greater financial incentives to file for UI benefits; (ii) UI benefit recipients include a wider range of the unemployed than previous recessions including many underemployed workers; and (iii) the possible underestimate of unemployment suggested by the BLS appears much smaller than what has been reported.
The Possible Misclassification of Employees Absent from Work
The conflicting unemployment figures may be due to a possible misclassification of respondents to the Current Population Survey (CPS), the monthly household survey used to estimate the official unemployment rate. Not surprisingly, more survey respondents in the past few months indicated that they were employed but absent from work. Interviewers were told to classify employees who were absent from work because their employer had shut down from the COVID-19 pandemic as being on temporary layoff (a BLS unemployment category). The CPS reported enormous increases in the number of workers on temporary layoff, but also found quite large increases in the number of employees absent from work for “other reasons.” These absent employees, like those with paid time off for vacation or sick leave, are counted as employed despite being absent from work.
In an abundance of caution the BLS reported this possible misclassification and indicated that if the entire increase in employees absent from work due to “other reasons” was instead a surge in temporary layoffs, the number of unemployed workers would have been higher by about 1.6 million in March, 7.6 million in April, and 4.7 million in May.1 Figure 1 illustrates the substantial adjustments that would be made to the number of unemployed workers if all new absences for “other reasons” were considered layoffs and not paid time off.2
One way to assess the extent to which unemployment may have been underestimated in the CPS is to compare CPS estimates to estimates from the Current Employment Statistics (CES) establishment survey. While these two surveys are not directly comparable, the BLS produces a CPS research series that can be compared to the establishment survey.3 Because the potential temporary layoff underestimate described above leads to an equivalent but opposite overestimate of employment, we adjust the CPS research series for the potential misclassification of absent workers and compare recent changes in both household (CPS) and establishment (CES) employment.
Establishment respondents to the CES were told to include employees who were absent from work but receiving salaries in their employment totals. Absent workers not receiving salaries were to be excluded from employment totals even if they received benefits. Figure 2 shows the comparison of the CES employment figure with the unadjusted and adjusted CPS employment figures. If workers in the questionable absent from work for “other reasons” category were paid salaries and correctly classified as not laid off, the unadjusted household (CPS) estimates of employment should be similar to the CES employment data. However, if workers absent from work for “other reasons” were actually laid off (and not paid), the adjusted household (CPS) employment estimates should be similar to the CES employment data.
The reclassification of absent workers from employed to unemployed, as suggested by the BLS, creates an even larger divergence between estimated employment declines in the CPS household and CES data. Because the CES employment estimate is closer to the unadjusted CPS estimate than the adjusted CPS estimate, we tentatively conclude that it is possible that no misclassification occurred among CPS workers who responded they were absent from work for “other reasons.”
Comparing Unemployment to the Number of Workers Collecting Unemployment Insurance
Since 1967, when UI benefit recipient data was first reported, there have never been more workers receiving UI benefits than the official estimate of unemployed workers. Historically, the number of workers collecting UI benefits, which is based on administrative data from state agencies, is much lower than estimated unemployment from the CPS. This is because only job losers qualify for UI while unemployed “job leavers,” new entrants, or re-entrants to the labor force do not qualify for UI. Self-employed workers and workers in the gig economy do not pay into the UI system and are also typically ineligible for UI, not all job losers are eligible for UI due to minimum prior work and earnings requirements, and not all eligible unemployed workers file for UI benefits.
For example, during March to May of 2019 the number of UI recipients was, on average, about 67.1% of unemployed job losers, and 30.6% of all unemployed workers. In recessions, the share of the unemployed who are job losers tends to increase so that the ratio of UI recipients to unemployment increases. During the Great Recession, the Federal government substantially increased UI benefits so that in many months more than half of UI recipients were receiving benefits from special Federal programs. As a result, from 2009 to 2011 the ratio of UI recipients to unemployed job losers was 99.5% and the ratio to all unemployed workers was 61.6%.
Figure 3 illustrates the weekly increases in the number of UI recipients compared to changes in the estimated number of unemployed workers and unemployed job losers in the CPS. In the past few months, the ratio of UI recipients to the number of unemployed workers has been quickly rising and in May reached 147%. The ratio of UI recipients to unemployed job losers was even higher, at 169%. Even if the official unemployment totals in May are adjusted to include all workers absent from work for “other reasons,” the ratio of UI recipients to the unemployed is 119%.
Many UI Recipients are Underemployed, Not Jobless.
Part of the reason for the large increase in UI recipients since March is the passage of the Coronavirus Aid, Relief, and Economic Security (CARES) Act. The CARES Act has expanded UI coverage to include the self-employed, workers in the gig economy, and workers with more limited work histories who would otherwise have been ineligible for UI benefits. While many states allow furloughed workers or employees with a substantial reduction in work hours to receive UI assistance, the CARES Act also provides greater financial incentives for these underemployed workers to file for UI benefits by granting an additional $600 per week to all UI recipients until July 31. Thus, while most states limit UI benefits to a worker with reduced hours, so that the total of earnings plus benefits do not exceed full UI benefits, even workers with a relatively small regular UI benefit payment will collect an additional $600 per week from the Federal Government.
How many workers are there on furlough or with reduced work hours due to the COVID-19 pandemic? The answer appears to be at least 11 million. In addition to the roughly 4.7 million excess workers absent from work for “other reasons,” in May there were 6.26 million more part-time workers due to economic reasons compared to May 2019. If these 11 million underemployed workers are added to the official total of 20.5 million unemployed workers, the 30 million UI recipients in mid-May would equal about 95% of the unemployed plus underemployed due to the pandemic.
The economic downturn caused by the pandemic has resulted in an unprecedented difference between the number of UI benefit recipients and the number of unemployed workers. Despite concerns raised by the BLS that workers on temporary layoff may have been underestimated and misclassified as employed but absent from work, our comparison of employment estimates from establishment and household surveys suggests otherwise. A simpler explanation for the gap between UI recipients and estimated unemployment is that the CARES Act has provided a much greater financial incentive for underemployed workers to apply for UI benefits. The surge in UI recipients may be because even underemployed workers, who would normally receive minimal assistance for a reduction in income and work hours, receive an additional $600 per week in Federal benefits until July 31.
- The past three Employment Situation Reports have indicated in the technical notes that had these adjustments been made, the unemployment rate would have been about 5.5% in March, 19.3% in April, and 16.0% in May (rather than 4.5%, 14.4%, and 13.0% respectively).
- Employees on temporary layoff are typically a small share of unemployed workers. From 1967 to 2019 workers on temporary layoff accounted for 13.4% of the unemployed, on average, and in March-May of 2019 temporary layoffs accounted for 12.7% of the unemployed. The adjustments suggested by the BLS would increase the current reported share of temporary layoffs among the unemployed in April from 79% to 84% and in May 73% to 78%.
- The research series excludes the self-employed and gig economy workers, unpaid family workers, farm workers, and household workers such as nannies. https://data.bls.gov/timeseries/LNS12032187&years_option=all_years