On Friday, the Bureau of Labor Statistics (BLS) will release September’s numbers on the state of the U.S. labor market. August employment growth came in lower than June or July due in part to the Delta variant spreading quickly. Another sign of weakness in the August job report was the rise of Black unemployment, which remains significantly higher than other groups. The August jobs report showed us, once again, just how much the ebbs and flows of the pandemic are the dominant influence by far on trends in the labor market.

This pronounced August slowdown also came after more than half of all states prematurely ended the pandemic unemployment insurance (UI) programs. All of the pandemic programs ended in early September.

While no longer accelerating, COVID-19 caseloads were still high in mid-September compared with the early summer months, so that may once again slow the recovery. In this jobs report, one indicator I’m going to be paying close attention to is public-sector employment, specifically public-sector education employment.

Overall government employment is still down 790,000 jobs since February 2020—the third largest employment deficit in any economic sector—but this shortfall is entirely in state and local employment and much of those losses were in education employment. Local education employment—think the public K-12 school system—cratered in the spring of 2020, experiencing losses in excess of those suffered in the Great Recession and, as of mid-August, remained 219,600 short of its pre-pandemic level.

In late August and early September, many schools across the country reopened their doors to in-person schooling five days a week for the first time since March 2020. As a result, I expect a significant pickup in local education employment.

However, there is a catch. Every year, education employment spikes in September, but seasonal adjustment factors instituted by BLS moderate those gains in order to remove seasonality. This makes sense—the same thing happens with retail employment in December. That’s when retail employment tends to spike in response to holiday shopping, and seasonal adjustments remove that bump so analysts can better ascertain the underlying trends. This is also why when the weather is unseasonably bad (or good) employment may decrease (or increase) more than expected in a usual season.

Here’s how this phenomenon plays out in public K-12 employment. In the figure below, if you look at 2017 through 2019, you can see how the red line (seasonally adjusted series) offsets the cycles of the blue line (nonseasonally adjusted series) to yield the underlying longer-term trends of education employment slowly rising year after year. The nonseasonally adjusted series (NSA) can be considered the actual level of employment in that sector, while the seasonally adjusted series (SA) is the one that gets more often reported for its consistency across months within a year. Eyeing the nonseasonally adjusted series, you can see a large dip in the summer months and a smaller dip in January for each of those years as you would given the typical school year.

Enter the pandemic. In spring 2020, both the seasonally and nonseasonally adjusted series fell sharply. But in September 2020, nonseasonally adjusted employment rose quickly, though the seasonally adjusted series reported a loss because it didn’t rise as quickly as in normal years. The January 2021 dip wasn’t as pronounced as in prior years, so the seasonally adjusted series reported a small gain. In June and July 2021, employment fell, but not as fast as usual, so the seasonally adjusted series reported gains.

What will happen with public K-12 employment in September? If 2020 is a guide, it is likely that employment will increase dramatically, but if employment is not as fast as “expected,” the seasonally adjusted series could report a loss or very little change in either direction. This may also depend on how large the September 2020 experience plays in the seasonal adjustment factor itself.

In 2020, the failure of local education employment to see its normal seasonal bounce was unambiguously because demand for these employees was depressed. Remote school, for example, meant that fewer bus drivers and cafeteria workers were needed. Currently, however, labor supply might be a constraint, given recent reports of difficulty in hiring for certain school system positions such as bus drivers, food service workers, building maintenance workers, or substitute teachers.

These potential hiring difficulties are not surprising given the ongoing pandemic and the fact that these are lower wage and often part-time positions that employ an often-older workforce that may be more concerned about infection than others. Policymakers serious about alleviating labor supply constraints (and not just interested in ending UI programs) would institute safety protocols—like requiring masks to be worn whenever community spread of the virus is significant—to make these education support jobs safer for their traditional workforce.

When the data come out on Friday, we can track nonseasonally and seasonally adjusted employment, and better measure the shortfall in education employment since the pandemic began—on top of the shortfall that never closed in the aftermath of the Great Recession.