2021 Workforce Data Revisions: Job Losses Were Not Quite as Sharp as Previously Indicated

Each month estimates from two surveys provide the most current indications of workforce conditions. The Current Population Survey, often referred to as the “household survey,” is the primary source for estimates of unemployment and labor force participation of the population. The Current Employment Statistics “payroll survey” is the source for estimates of the number of jobs in many nonfarm industries. 

In March of each year, estimates from those programs are revised based on lagging but more complete data. Compared to previously published estimates, revised data for 2019 and 2020 indicates that:

• Unemployment was lower for most months
• Labor force participation was higher in 2019 and early 2020, before the pandemic disrupted labor market activity, and mostly lower for the remainder of the year
• The number of nonfarm payroll jobs was slightly higher in 2019 and early 2020, and significantly higher for much of the rest of the year

Data cited in this brief is seasonally adjusted. 

Unemployment Rate

Previously published estimates indicated that the unemployment rate was nearly unchanged in 2019 and the first three months of 2020, ranging between 2.9 and 3.2 percent. It shot up to a record 10.4 percent in April after the pandemic began to impact the labor market. In the following months unemployment rates moved sharply and erratically, down, up, and back down, before settling into a more consistent, gradual downward pattern in the fall. The variability of direction and the magnitude of monthly change that occurred between March and October was without precedent. 

Revised rates indicate that unemployment was even lower in 2019 and early 2020, ranging from 2.6 to 3.1 percent. Unemployment reached a high of 9.1 percent in April, now officially the highest rate since February 1977. Rates remained historically high and moved erratically throughout the summer but were an average of 1.3 percentage points lower than previously published estimates indicated for the period from April to September 2020. Unemployment rates settled close to five percent from August until year-end. 

Unemployment rate revisions for the spring and summer months are much larger than usual -- likely the largest that have ever occurred. The pandemic disrupted labor market activity with business closures and reduced spending on many services. It also disrupted normal survey operations. The U.S. Bureau of Labor Statistics reports that response rates to the household survey were lower than normal during that period. Additionally, there was a great deal of uncertainty for many people about their status as to whether they were temporarily or permanently laid off from a business, not knowing when or if it may reopen.

Labor Force Participation

Previously published estimates indicated that labor force participation was little changed in 2019 and the first three months of 2020, ranging between 62.3 and 62.7 percent. In April it decreased sharply to 59.1 percent before recovering gradually in May and June. In the following months it surged to pre-recession levels in July before sinking through most of the rest of the year. Like unemployment estimates, the variability of direction and the magnitude of monthly change that occurred between March and October was without precedent. 

Revised rates indicate that participation in the labor force was somewhat higher in 2019 and early 2020, and somewhat lower in all but one month from April to the end of the year. Participation rates shot up and down each month by as much as 3.9 percentage points and an average of 2.5 points each month between March and September. Changes of this magnitude are unusual for an entire year and have never occurred for a single month, let alone in six consecutive months. 

Several factors contributed to the variability of labor force participation and unemployment rates that began in April 2020. Business, school, and daycare closures and re-openings, as well as stay-at-home orders and health concerns impacted whether many people had a job or whether they could be available for work. Uncertainty over how long the pandemic would impact workplaces left many people unsure whether they had a job to return to or whether they should look for a new job. The largest factor likely is that response rates to the survey were much lower during the pandemic and that the statistical models used to produce these estimates were not designed to handle this once in a century situation. The gyrations in published labor force participation rates in the spring and summer likely are not indicative of what actually occurred in terms of human behavior.

Nonfarm Payroll Jobs

Preliminary estimates indicated the number of nonfarm payroll jobs had a relatively consistent pattern of growth in 2019. The pandemic caused a precipitous decrease of 104,500 jobs between February and April, followed by a strong rebound in May and June, then gradually smaller gains through the end of the year. Job losses in the spring impacted nearly every sector of the economy.

Revised estimates indicate there were somewhat more jobs in 2019 and early 2020 than previously indicated, and an average of 10,000 more jobs in the spring. Between February and April the net loss was 94,600 jobs. (This is larger than for any period since at least the 1940s, and likely larger than for any years or stretch of years during the Great Depression of the 1930s). The year ended more positively than previously thought, as well. The number of jobs for the period from September to December was revised up an average of 12,500. Though there were more jobs at the end of the year than previously indicated, there was markedly slower growth during the fall because of the virus surge. 

The Paycheck Protection Program was passed by Congress shortly after the pandemic caused many businesses to temporarily close. It provided forgivable loans to employers to keep staff on their payroll. Because those funds came through months after many had been laid off, thousands of people effectively were put back on payrolls retroactively. This is likely a significant factor in the magnitude of upward revisions to jobs data in the spring. 

The number of nonfarm jobs increased each year from 2010 to an all-time high of 637,100 in 2019. In 2020 there was an average of 41,000 fewer jobs. Losses were spread across many industries, with the largest in the leisure and hospitality, education (public and private K-12 and higher ed combined), retail trade, health care and social assistance, and manufacturing sectors. The number of jobs in the construction and mining and logging sectors were nearly unchanged in 2020, and federal government added an average of 800 jobs. This was partially from temporary jobs related to the decennial Census of Population and partially from hiring at the Portsmouth Naval Shipyard in Kittery.

Annual revisions improve the quality of workforce data and provide a better indication of overall trends. We expect variability in preliminary estimates to continue through 2021 and beyond. This is due to a number of factors, including changes in the sample of survey respondents and to statistical methods that are used. As has occurred in recent years, labor force and unemployment estimates are likely to move in one direction for several months before reversing course in the other direction. Directional turns in unemployment rates that may run for three to six months often are not indicative of a fundamental change in underlying workforce conditions. As 2021 progresses, we recommend looking at trends over many months to draw conclusions about the direction and state of the labor force.