13th annual Municipal Finance Conference


13th annual Municipal Finance Conference


Is ‘long Covid’ worsening the labor shortage?

Cover photo

Over the past two years, the media, politicians, public health authorities, and others have paid close attention to the number of lives lost to COVID-19: more than 800,000 Americans so far. Yet we hear much less about a second number: the Americans who have lost their health to COVID-19.

Millions of COVID-19 patients have developed a range of debilitating symptoms that last for months or even years. They are being diagnosed with “post-acute sequelae of COVID-19”—or more colloquially, long Covid. Yet we know little about these people—how many there are, why they stay sick, or what the impact is on their lives. Among these knowledge gaps is the fact that public health and economics experts have almost no understanding of long Covid’s economic burden.

This piece explores data suggesting that long Covid is contributing to record high numbers of unfilled jobs by keeping millions of people from getting back to work.

Long Covid could account for upward of 15% of unfilled jobs

With 10.6 million unfilled jobs across the country, the months-long labor shortage is weighing on the U.S. economy. Small businesses are losing money due to understaffing. Local governments are struggling to fill jobs. Investors are spooked, and corporate profits are taking a hit.

Economists have proposed a number of explanations, including a decline in workers’ willingness to tolerate low pay and poor working conditions, lack of access to child care, concerns about contracting COVID-19, higher household savings, and demographic and immigration trends. Yet they rarely mention long Covid.

The Centers for Disease Control and Prevention estimates that through October 2021, just over 100 million Americans between the ages of 18 and 64 have contracted COVID-19. And studies suggest that between 27% and 33% of COVID-19 patients still experience symptoms months after infection. That means 31 million working-age Americans—more than one in seven—may have experienced, or be experiencing, lingering COVID-19 symptoms.


Figure 1

It may not be the case that all 31 million are still sick; some may have recovered. Understanding how many people have long Covid at any given time requires an assumption about average illness duration. In the U.K., which is doing a much better job collecting data than the U.S., more than 70% of people with persistent COVID-19 symptoms have been sick for more than three months, and more than one-third have been sick for at least a year. This chronicity is consistent with other post-viral illnesses, which behave similarly to long Covid and often last for years.

To be conservative, this analysis assumes the 31 million long Covid patients stayed sick for an average of three months. That means that about 4.5 million may have been sick at any given time over the past 20 months.

Not all of these 4.5 million people would have stopped working. Two studies of long Covid patients found that 23% and 28%, respectively, were out of work due to long Covid at the time of the study. That suggests there may have been about 1.1 million Americans not working due to long Covid at any given time.

Additionally, some long Covid patients reduced hours rather than taking time off: 46% according to a study in The Lancet. That is another 2.1 million workers. If those workers reduced their hours by only a quarter, that would increase the labor market impact to 1.6 million full-time equivalent workers. In other words, under reasonable assumptions given the data available, long Covid could account for 15% of the nation’s 10.6 million unfilled jobs.

Figure 2

Long Covid is rarely part of labor shortage discussions, potentially due to inadequate data

The above estimate requires several assumptions, and so may be proved inaccurate. But the magnitude of the numbers involved suggests that long Covid merits consideration in discussions about the labor shortage. So why don’t we hear more about it? Poor data may be the culprit; there is no clear and comprehensive source of data about how many people are not working due to long Covid.

Consider the two most likely sources for such data. The first is the Census Bureau’s Household Pulse Survey (HPS), which provides quick-turnaround data on COVID-19’s impact on Americans’ lives. The 20-minute virtual survey asks respondents their employment status, and if they are not working, why. The two responses most relevant to long Covid are “I am/was sick with Coronavirus symptoms” and “I am/was sick (not coronavirus related) or disabled.”

Long Covid does not fit well into either response. Respondents may assume they need an active infection to respond “yes” to the first; further, many long Covid symptoms do not overlap with those of an active infection. The second response specifies that the reason is not related to coronavirus, which long Covid obviously is. (Note that more than 20% of respondents check “other” as their reason for not working.)

The second data source is the Current Population Survey (CPS), a monthly in-person survey of 60,000 U.S. households conducted by the Census Bureau for the Bureau of Labor Statistics (BLS). It is the central source of information on U.S. labor force statistics, counting unemployed Americans (not working but actively looking for work) and those not in the labor force (not working and not looking for work). As with the HPS, it is not obvious where long Covid fits     . Workers disabled by long Covid could fall into either category or neither; for example, they could be employed but on unpaid sick leave from work.

Further, CPS respondents can choose “ill health” as their reason for not working only if: (1) they are out of the labor force now; (2) they want a job; (3) they searched for a job in the last year; but (4) did not search for a job in the past four weeks. Yet someone with long Covid might not want a job now due to health constraints, and they might not have searched for a job over the past year because they might have been employed until they got sick. (In May 2020, the CPS added five Covid-specific questions, but none ask about inability to work due to COVID-19, long or otherwise.)

Better long Covid data could inform disability policy, public health guidance, medical research funding prioritization, and more

To fill this data gap, the Census Bureau and BLS should introduce questions about long Covid disability to both the HPS and CPS. As Lisa McCorkell, co-founder of the long-Covid-focused Patient-Led Research Collaborative, told Brookings, “Until we have data from a representative sample that accurately capture the extent of the impacts to the labor force, economists and policymakers are likely not going to consider long Covid an economic issue or recognize it for the mass disabling event it is.”

The HPS’ advantage is speed. Because it is part of the Census Bureau’s Experimental Data Series, the bar for statistical quality is lower, which may enable faster change. The CPS, meanwhile, has a higher sample size, rigor, and reliability, and should be used to track long Covid’s impact alongside the HPS.

Crafting the right questions about long Covid is critical. To do so, the Census Bureau and BLS should work with National Institutes of Health teams researching long Covid as well as long Covid patient advocacy groups—these patients are best placed to identify trends and potential data pitfalls that survey designers might otherwise miss. Questions should focus on the information policymakers need most, including:

  • The number of full-time equivalent workers currently not working due to long Covid (including those at reduced hours)
  • Average time off or time spent at reduced hours due to long Covid
  • Workplace accommodations that would enable long Covid patients to increase working hours
  • Applications, approvals, and rejections for Social Security Disability Insurance among long Covid patients

Access to this data will help policymakers and businesses better predict how today’s labor market conditions may evolve. It may also prompt the government to improve disability standards so that people with long Covid can request accommodations or secure benefits; issue health guidance that takes into account the economic burden of large-scale disability; and provide more federal dollars for long Covid research and medical care. The possibilities are vast, but first, we need the data.


  • Footnotes
    1. Estimates of symptom persistence range. A recent meta-analysis found that three months after diagnosis, more than 30% of COVID-19 patients experience fatigue, and more than 20% experience cognitive impairment. Studies from Imperial College and the University of Oxford found that about one-third of COVID-19 patients experience symptoms months after infection. A study from the U.K. Office for National Statistics put symptom prevalence at 14% one month after infection. Another recent meta-analysis found that that number may be closer to 50%.
    2. A widely accepted long COVID definition from the World Health Organization (WHO) requires that the illness last for at least two months. In this piece, “long Covid” is defined as lingering symptoms after COVID-19 infection, regardless of duration. The average illness duration of a patient meeting the WHO criteria will probably be longer than three months.