COVID Relief Spending

Introduction

Beginning in FY2020, the F-33 survey added items tracking district expenditures of temporary federal COVID-19 assistance funds (ESSER and related programs). edfinr’s full dataset carries these as the exp_covid_* variables, which let you see how much pandemic aid districts reported spending, on what, and how spending ramped up and began to decline from FY20-FY23.

library(edfinr)
library(dplyr)
library(tidyr)
library(ggplot2)

The eight edfinr COVID expenditure variables map to F-33 items AE1-AE8 (see list_variables("full")):

• exp_covid_total (AE1) – total expenditures from COVID-19 federal assistance funds.
• exp_covid_instr (AE2) – instructional expenditures.
• exp_covid_supp (AE3) – support services expenditures.
• exp_covid_cap_out (AE4) – capital outlay expenditures.
• exp_covid_tech_supp (AE5) – technology-related supplies and purchased services.
• exp_covid_tech_equip (AE6) – technology-related equipment.
• exp_covid_supp_plant (AE7) – operation and maintenance of plant (from FY2021).
• exp_covid_food (AE8) – food services operations (from FY2021).

Important caveats

1. Full dataset only. The exp_covid_* variables are not in the skinny dataset; request dataset_type = "full".
2. Availability. AE1-AE6 begin in FY2020; AE7 and AE8 begin in FY21 (first_yr_avail in the dictionary records this).
3. NA is not zero. Districts that did not report an item show NA, including districts whose raw F-33 values are zero-filled but flagged missing (FL_AE1 = "M"); edfinr converts those to NA during cleaning. New York did not report these items, so its districts are NA in every year FY2020-FY2023, and roughly a third to half of California districts are NA from FY2021 onward. A 0 that survives cleaning is a genuine reported zero, which is common in FY2020 when most ESSER funds were not yet spent. Treat state-level comparisons with care.
4. CPI adjustment applies. Like other expenditure flows, the exp_covid_* variables are scaled when you pass cpi_adj. The examples below use nominal dollars to match how federal awards are usually described.

The arc of relief spending

covid <- get_finance_data(yr = "2020:2023", geo = "all", dataset_type = "full")

national <- covid |>
  filter(!is.na(exp_covid_total), !is.na(exp_cur_total)) |>
  group_by(year) |>
  summarize(
    covid_total = sum(exp_covid_total),
    cur_total = sum(exp_cur_total),
    enroll = sum(enroll, na.rm = TRUE),
    .groups = "drop"
  ) |>
  mutate(
    covid_pp = covid_total / enroll,
    covid_share = covid_total / cur_total
  )

national |>
  mutate(
    covid_total_b = round(covid_total / 1e9, 1),
    covid_pp = round(covid_pp),
    covid_share = round(100 * covid_share, 1)
  ) |>
  select(year, covid_total_b, covid_pp, covid_share)

## # A tibble: 4 × 4
##    year covid_total_b covid_pp covid_share
##   <int>         <dbl>    <dbl>       <dbl>
## 1  2020           1.9       51         0.4
## 2  2021          22.7      532         4.1
## 3  2022          33.9      808         5.6
## 4  2023          35.7      819         5.3

The table shows total reported COVID-fund spending (in billions), per-pupil spending, and COVID-fund spending as a share of current expenditure among districts reporting both items.

What did districts spend relief funds on?

component_labels <- c(
  exp_covid_instr = "Instruction",
  exp_covid_supp = "Support services",
  exp_covid_cap_out = "Capital outlay",
  exp_covid_tech_supp = "Technology supplies/services",
  exp_covid_tech_equip = "Technology equipment",
  exp_covid_supp_plant = "Plant operations",
  exp_covid_food = "Food services"
)

covid |>
  select(year, all_of(names(component_labels))) |>
  pivot_longer(-year, names_to = "component", values_to = "amount") |>
  filter(!is.na(amount)) |>
  group_by(year, component) |>
  summarize(total = sum(amount), .groups = "drop_last") |>
  mutate(share = total / sum(total)) |>
  ungroup() |>
  mutate(component = component_labels[component]) |>
  ggplot(aes(x = year, y = share, fill = component)) +
  geom_col() +
  scale_y_continuous(labels = scales::label_percent()) +
  labs(
    title = "Composition of Reported COVID-Fund Spending by Year",
    subtitle = "Shares of summed component items (AE2-AE8), nominal dollars",
    x = "Year", y = "Share of Component Spending", fill = "Component"
  ) +
  theme_minimal()

How did relief spending vary across districts?

covid |>
  filter(year == 2023, !is.na(exp_covid_total), enroll >= 200,
         !is.na(urbanicity)) |>
  mutate(covid_pp = exp_covid_total / enroll) |>
  ggplot(aes(x = urbanicity, y = covid_pp)) +
  geom_boxplot(outlier.alpha = 0.15) +
  scale_y_continuous(labels = scales::label_dollar()) +
  coord_cartesian(ylim = c(0, 5000)) +
  labs(
    title = "Per-Pupil COVID-Fund Spending by Urbanicity, FY2023",
    subtitle = "Districts with 200+ students; y-axis truncated at $5,000",
    x = "Urbanicity", y = "COVID-Fund Spending Per-Pupil"
  ) +
  theme_minimal()

Tracking the wind-down in large districts

largest_ids <- covid |>
  filter(year == 2023) |>
  slice_max(enroll, n = 6) |>
  pull(ncesid)

covid |>
  filter(ncesid %in% largest_ids) |>
  mutate(
    dist_name = dist_name[which.max(year)],
    covid_pp = exp_covid_total / enroll,
    .by = ncesid
  ) |>
  ggplot(aes(x = year, y = covid_pp)) +
  geom_line() +
  geom_point(size = 1) +
  facet_wrap(~dist_name, ncol = 2) +
  scale_y_continuous(labels = scales::label_dollar()) +
  labs(
    title = "Per-Pupil COVID-Fund Spending in the Six Largest Districts",
    subtitle = "FY2020-FY2023, nominal dollars",
    x = "Year", y = "COVID-Fund Spending Per-Pupil"
  ) +
  theme_minimal()

The empty New York City panel is the sharpest illustration of caveat 3’s warning: the nation’s largest district shows NA on every AE item in all four years. The raw F-33 files carry those items as zeros, but the accompanying data-item flags mark them missing (FL_AE1 = "M") – New York State never reported the COVID items for any of its districts – so edfinr labels them as NA rather than letting billions of dollars in allocated relief funds appear as zero spending. Before reading any single district’s series, confirm the state actually reported these items in the years you are comparing.

See also

• The “Data Quality and Comparability” article for missingness and reporting caveats that apply doubly to these items.
• The “CPI Adjustments” vignette if you need relief spending in constant dollars.