Last year I wrote a wonky deep dive into how we count the number of workers in schools.
I want to follow up on that piece by adding another year’s worth of data from the Census Bureau’s Annual Survey of Public Employment and Payroll (ASPEP). The data now extend through March of 2023, and I created a simple graph to show the problem.
The chart below has two lines. The light green line is measuring the total number of employees in K-12 schools. It adds up all full-time and part-time staff to arrive at one total. You can think of this as a headcount—it’s literally just counting each employee as one head, regardless of how many hours they work per week. This is a fast and easy way to measure employment, and it’s also how the BLS reports its monthly jobs numbers.
But there’s another way to measure employment levels called “full-time equivalents,” or FTEs. It’s the red line in the graph below.
It takes longer to calculate FTEs and to collect the data for every school and district in the country, but it’s a more sophisticated count of total employment. To calculate FTEs, you need to know how many full- versus part-time workers there are, and how many hours the part-timers work each week.
As you can see in the chart, these two lines moved pretty much in tandem for most of the early 2000s, but they started to pull apart about a decade ago.
More recently, the two ways of measuring employment told very different stories about school staffing levels in the wake of the pandemic:
Schools lost about 435,000 total employees from March 2020 to March 2021. But, because almost all of these were part-time staff, the change in FTEs was only half that amount.
While the total employee count was still lower in March 2023 than it was in March 2020, the FTE count had fully recovered by 2022 and was hitting all-time highs in 2023.
The headcount employment numbers returned to their pre-pandemic levels late last year, so FTE counts in schools are likely higher now as well.
Personally, I prefer the FTE count as a better measure of total staffing levels, but it’s important to understand what each data point can and cannot tell us, especially because they have diverged in important ways over the last few years.