New Evidence Confirms Long Covid Is Indeed Dragging Down LFP. NewsWire

NewsWire: 9/26/22

• A new study estimates that workers who take a week off work for Covid-19 are 7% less likely to be in the labor force a year later. The researchers conclude that Long Covid has reduced the U.S. labor force by at least 500K people--an extremely conservative estimate. (National Bureau of Economic Research)
• NH: In July 2021, we wrote our first article arguing that Long Covid was partly responsible for America's low LFP rate. (See “Long Covid Casts a Long Shadow.”) At the time, few others were making this claim. More recently, the persistent LFP deficit has persuaded others that the claim makes at least some sense. Now a new study published by the NBER, using an innovative method of analyzing BLS worker surveys, comes to a similar conclusion. So let's explore the paper. 
• The main challenge in studying the effects of Long Covid is the lack of any firm data. Because there are no clear, agreed-upon criteria for diagnosing Long Covid, we cannot clearly define the population who is actually suffering from it. We're not even sure exactly who has or has not contracted the acute form of Covid-19, since a lot of people never run a PCR or antigen test while they're sick or an antibody test after they recover. To get around such hurdles, this team of economists employs an indirect method of identifying the link between acute disease and later disability.
• The researchers begin by looking at the absentee section of the Current Population Survey (CPS), a monthly poll of households conducted by the Census and BLS. The survey marks workers as "absent" if they worked zero hours in a reference week. Respondents are then asked why they did not work. Of the fourteen possible answers, the researchers focused on missed work due to "own illness/injury/medical problems." (We have reported on similar CPS data in the past--see “Reasons Americans Aren't Working.”)
• In the decade before the pandemic (January 2010-February 2020), six workers per thousand on average missed a full week of work for health reasons. During the pandemic (March 2020-June 2022), this share rose to ten per thousand. Using national and local Covid-19 case numbers, the researchers found that swings in excess absenteeism largely mimicked swings in case counts. They concluded that the excess number of absent workers serves as a reasonable proxy for severe Covid-19 illness. 
• The researchers then tracked these probable Covid-19 survivors' labor force status during subsequent months. It turns out that, after 12 months, the survivors were indeed more likely than other workers to have left the workforce.
• To make sure that they weren't following workers who were already sicker before the week-long absence, the researchers imposed two sets of controls. First, they controlled for demographic differences between the absent and non-absent groups (age, sex, earlier history of absences, etc.) And second, they controlled for any tendency for absent workers to quit the labor force a year later that may have been present before the pandemic began in 2020.
• The results? After imposing both sets of controls, the researchers found that, compared to workers who did not have a health-related absence, these severe Covid-19 survivors were -7% less likely to be in the labor force a year later. (These rates are similar "to the effects of hospitalizations on participation" in the pre-pandemic era.) And, because many of the absent workers later work fewer hours, even if they still remain in the workforce, their reduction in dollar earnings is greater: about -18% less.

• Extrapolating these results, the authors ultimately conclude that Covid-19 illness has reduced the LFP rate by -0.2 percentage points. That translates to roughly 500K people currently not working. The researchers then add that this is a conservative estimate. The CPS doesn’t count people who were not in the labor force but may have joined if they hadn't gotten sick. Nor does it count case infection surges during the weeks the survey didn't run. Accounting for these variables, they conclude that the reduction in the LFP rate may be closer to -0.3 percentage points, or 750K people not working.   
• While the researchers are right to account for any week-long absences that slipped through their net, they don't seem to be aware of the full implications of their findings. Their study strictly limits itself to workers who reported being absent from work for an entire week. If 7% of them amounted to 500K, then the total number of week-long absent workers was only 7M.
• Hmmm. Over the past couple of years, there have been 160-165M workers in the labor force. So are the researchers implying that only 4-5% of these workers suffered from acute Covid-19%? That seems hard to believe. They must know that the actual share is much larger--roughly two-thirds, similar to the infected share for the entire U.S. population. The researchers may be assuming that only severely ill patients are at risk for Long Covid. Yet we know from abundant research that Long Covid shows little correlation with age or severity of illness. Even those who are asymptomatic after testing positive have a 19% risk of Long Covid.
• In other words, while the researchers produce a very reasonable estimate of workforce dropouts from the week-long absent workers they study, those workers represent only a small fraction of all the infected workers who are at risk for Long Covid--most of whom, while acutely infected, chose not to be absent, were not sick enough to be absent, or indeed were entirely asymptomatic.
• If we apply the researchers' 7% dropout rate to the total number of workers who have been infected (probably around 100 million), we get a much larger number: roughly 7 million. Now I agree that 7 million is too high. There is, after all, some correlation between illness severity and Long Covid. And, I imagine, the sort of worker who, even if severely ill, would still come to work is probably the sort of worker who is less likely to leave the workforce a year later.
• What if we adjusted for these effects by reducing the researchers' 7% dropout rate, for the entire workforce, by one-third? That would generate a 4.7 million reduction in the workforce. In December, I presented a similar estimate: 4.3M workers. (See “The Impact of Long Covid Keeps Growing.”) We calculated this estimate by assuming, based on survey evidence, that 20% of the total number of Covid infections in the U.S. would result in Long Covid. We then calculated the share of these Long Covid sufferers who were likely to be in the labor force (≈66%). And finally we assumed, again based on survey evidence, that 22% of these were unlikely to be able to work.
• Put differently, our basic calculation from Long Covid to workforce dropout was 0.20 x 0.22 = 0.044. Thus, we found that 4.4% of workers who contract Covid-19 leave the labor force. Our percentage is somewhat more modest than the estimate of these researchers. But it's not far off. And we are starting with a much larger population of Long Covid sufferers: 19.5M vs 7.0M. 
• Though it fails (IMO) to generalize its conclusions to the entire workforce, I admire the NBER study for its clever data-sleuthing techniques. It's an impressive piece of work. Along the way, the study also provides some fascinating sidelights on the long-term sequelae of (presumed) short-term Covid infection by age of worker.

• Note that older Covid-19 survivors (over age 55 and especially over age 65) are indeed somewhat more likely to leave the workforce 1-2 months and 9-14 months after infection. While fewer than 5% of those under age 55 dropped out in 9-14 months, nearly 20% of those aged 65+ dropped out. This most likely reflects the decision of many older workers to retire after their severe illnesses.
• Now let's pan out and reflect on some of the broader macro implications of Long Covid.
• First, we should understand that there is nothing fundamentally new about Long Covid. History provides ample evidence that large viral pandemics are routinely followed by several years of debilitating post-viral syndromes.
• Following the "Russian Flu" of 1889-92 (probably H3N8, but possibly the first invasion of the OC43 common cold coronavirus), doctors reported an ongoing litany of post-viral conditions that we would today call "myalgic encephalomyelitis" (i.e., chronic fatigue syndrome). These included "neuralgia, neurasthenia, neuritis, nerve exhaustion, grippe catalepsy, psychosis, prostration, inertia, and anxiety." Following the Spanish Influenza of 1917-19 (almost certainly H1N1), similar reports continued into the late 1920s. These included a paralyzing form of Parkinson's Disease called "encephalitis lethargica" whose elderly catatonic sufferers were the subject of Oliver Sacks' best-selling 1973 book, Awakenings. Other reports have followed smaller post-World War II flu pandemics.
• My point is not to be alarmist, but simply realistic. We already know that, for at least the first year after infection, Covid-19 survivors have a higher mortality rate than demographically identical non-survivors. (See "The Long Shadow of Long Covid.") Certain syndromes that often show up after Covid 19 infections, such as Parkinson's Disease and Type 1 Diabetes, are already suspected of being linked to Covid-19. It's happened before after pandemics, and it is happening again. It's just that this time we are able to track the effects in much greater quantitative detail.
• Second, we must appreciate how Long Covid has warped our macroeconomic outlook. One key reason the Fed, along with so many other observers, underestimated the inflation threat back in 2021 and early 2022 was that it looked at the labor market and kept seeing plenty of slack. Their reasoning seemed sound: If the economy were running anywhere near full employment, why was the labor force still missing millions of Americans who had just been there at the beginning of 2020? We now know the error of their reasoning. They weren't accounting for the impact of Long Covid.

• Just last month, in August, the civilian labor force finally grew back to its pre-pandemic size: 164.7M people. Yet the population of age 16+ Americans has meanwhile continued to grow. If we apply the pre-pandemic February 2020 LFP rate (63.4%) to this month, we are still missing about 2.8M workers. That's remarkable, considering that the number of Americans not working due to Long Covid is almost certainly greater than 2.8M.
• What that tells me is that the labor market, along with wage growth, today continues to be a lot hotter than it was in February 2020. The CBO contributes to this distortion by continuing to insist, in every quarter since the recovery from the pandemic recession, that "actual" GDP remains stuck below "potential" GDP. Excuse me, but if strong nominal wage acceleration plus a record quit rate and a near-record unemployment rate is not a sign of actual exceeding potential... then haven't these terms lost any recognizable meaning?
• There may also be some hints here about the likely future course of Fed policy. I think the Fed is finally figuring this game out. Powell is no longer looking at the number of workers. He's looking at wage and employment cost growth, along with core PCE trends. As long as these are untamed, he won't relent. Since there's a lot of energy left in hiring, Powell may stay in hiking mode for a while. It could get nasty.
• Third and finally, let's acknowledge the long-term drag that Long Covid is likely to exert on total employment and GDP growth over the next two or three years. So long as net immigration remains subdued, total employment growth in a no-more-than-full-employment economy will remain close to zero through 2024 or 2025.
• Even the CBO, which assumes no recession and an LFP recovery among older workers, currently projects the labor force to grow by an average of only 0.5% from 2022 through 2026. It may well be less than that. Which means that real GDP growth is going to depend primarily on rising labor productivity. Since Q2 2019, annual labor productivity has risen at an average annual rate of 0.7%; from 2007 to 2019, it rose at 0.9%. Let's face it: We may be stuck, at best, in the 1.0-1.5% real GDP growth range--again, assuming away the impact of any near-term recession.
• During a recent 60 Minutes interview, Biden declared that "the pandemic is over." Indeed, case numbers and deaths are nowhere near previous peaks. And most people have returned to their normal lives. But lest we forget, many are still suffering from the virus. And our macro judgment, along with our macro outlook, may still  be suffering as well. 

To view and search all NewsWires, reports, videos, and podcasts, visit Demography World. For help making full use of our archives, see this short tutorial.