Capital and Facilities

Introduction

edfinr includes capital, debt, and fund-balance variables to provide more inforation on school facilities funding.

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

The skinny dataset includes total capital outlay in dollars and per-pupil terms (exp_cap_total, exp_cap_total_pp). The full dataset adds the detailed components and the debt/fund-balance picture:

• Capital outlay components: exp_cap_construction, exp_cap_land, exp_cap_equip_instr, exp_cap_equip_other, exp_cap_equip_nonspec.
• Debt service: exp_debt_interest (interest only).
• Debt stocks: debt_lt_begin, debt_lt_issued, debt_lt_retired, debt_lt_end, debt_st_begin, debt_st_end.
• Fund balances (fiscal year end): fund_bal_debt_svc, fund_bal_bond, fund_bal_other.

Use list_variables("full", category = "debt") to see the debt and fund-balance entries, and list_variables("full", category = "expenditure") for the capital outlay items.

Important caveats

Read these before ranking or comparing districts on capital spending.

1. Capital is not current spending. Capital outlay is excluded from exp_cur_total by definition. Do not add it to current spending to get a “total” without knowing what you are combining.
2. Capital spending is lumpy. A district that builds a school records a large exp_cap_total_pp in one year and near-zero in the surrounding years. Single-year per-pupil capital rankings are misleading; 3-5 year averages can provide a more accurate picture.
3. Compare per-pupil, not totals. Raw dollar rankings mostly rank district size. Only exp_cap_total_pp ships as a per-pupil column; divide the debt and fund-balance variables by enroll yourself, as the examples below do.
4. Bond-funded vs. pay-as-you-go. High capital outlay may be financed by borrowing rather than current resources. Read debt_lt_issued and fund_bal_bond alongside the outlay flows to understand how construction was paid for.
5. Interplay with the state-revenue adjustment. edfinr nets state capital and debt aid out of rev_state; that netted-out amount ships as rev_state_cap_debt, the pre-adjustment value is preserved in rev_state_unadj / rev_state_unadj_pp, and c11_spike_flag flags district-years where that adjustment spikes. Consider these when relating revenue to capital activity.
6. Stocks stay nominal. Debt and fund-balance stocks (debt_*, fund_bal_*) are balance-sheet levels and are never CPI-adjusted when you pass cpi_adj. Capital outlay flows and exp_debt_interest are adjusted.

Worked example: multi-year average capital per pupil

Because capital expenditures are lumpy, averaging over several years can produce more interperatable results. The example below examined Ohio school distircts, using CPI-adjusted 2023 dollars, to average per-pupil capital outlay over the five most recent years. Only districts with at least three years of data and at least 1,000 students are included, since very small districts can produce extreme per-pupil values from modest capital projects.

oh <- get_finance_data(
  yr = "2012:2023", geo = "OH",
  dataset_type = "full", cpi_adj = "2023"
)

cap_multiyear <- oh |>
  filter(year >= 2019) |>
  group_by(ncesid) |>
  summarize(
    dist_name = dist_name[which.max(year)],
    n_years = sum(!is.na(exp_cap_total_pp)),
    avg_cap_pp = mean(exp_cap_total_pp, na.rm = TRUE),
    avg_enroll = mean(enroll, na.rm = TRUE),
    .groups = "drop"
  ) |>
  filter(n_years >= 3, avg_enroll >= 1000) |>
  arrange(desc(avg_cap_pp))

head(cap_multiyear, 10)

## # A tibble: 10 × 5
##    ncesid  dist_name                 n_years avg_cap_pp avg_enroll
##    <chr>   <chr>                       <int>      <dbl>      <dbl>
##  1 3904500 Warrensville Heights City       5     13083.      1711.
##  2 3904560 Rossford Exempted Village       5     12939.      1613.
##  3 3904407 Grandview Heights Schools       5     11914.      1106 
##  4 3904524 Harrison Hills City             5     11392.      1452 
##  5 3904672 Northeastern Local              5     10930.      1039 
##  6 3904716 Berkshire Local                 5     10003.      1347.
##  7 3904493 Upper Arlington City            5      9994.      6221 
##  8 3904652 Wynford Local                   5      8449.      1159.
##  9 3910018 Warren Local                    5      8400.      2002.
## 10 3904780 Indian Creek Local              5      8271.      1978.

How lumpy is capital spending?

Plotting the full 2012-2023 series for Ohio districts with the highest recent averages shows why single-year rankings mislead: a district’s per-pupil capital outlay can swing by thousands of dollars from one year to the next as projects start and finish.

oh |>
  select(-dist_name) |>
  inner_join(
    head(cap_multiyear, 6) |> select(ncesid, dist_name),
    by = "ncesid"
  ) |>
  ggplot(aes(x = year, y = exp_cap_total_pp)) +
  geom_line() +
  geom_point(size = 1) +
  facet_wrap(~dist_name, ncol = 2) +
  scale_x_continuous(breaks = seq(2013, 2023, 4)) +
  scale_y_continuous(labels = scales::label_dollar()) +
  labs(
    title = "Per-Pupil Capital Outlay Is Lumpy",
    subtitle = "Ohio districts with the highest 2019-2023 average, in 2023 dollars",
    x = "Year",
    y = "Capital Outlay Per-Pupil"
  ) +
  theme_minimal()

How does capital’s share of spending vary by state?

Statewide aggregates smooth out district-level lumpiness and show how much of each state’s total K-12 spending goes to facilities. Shares are computed only from districts reporting both current and capital expenditure. Smoothing is not a cure-all: a state’s single-year share still moves with bond program cycles and could be affected by a single large district’s projects. Treat the ranking below as a snapshot; pool several years of data before drawing conclusions about a state’s ranking.

us_2023 <- get_finance_data(yr = "2023", geo = "all", dataset_type = "full")

state_cap <- us_2023 |>
  filter(!is.na(exp_cap_total), !is.na(exp_cur_total)) |>
  group_by(state) |>
  summarize(
    cap_share = sum(exp_cap_total) / (sum(exp_cur_total) + sum(exp_cap_total)),
    .groups = "drop"
  )

ggplot(state_cap, aes(x = cap_share, y = reorder(state, cap_share))) +
  geom_col() +
  scale_x_continuous(labels = scales::label_percent(accuracy = 1)) +
  labs(
    title = "Capital Outlay as a Share of Current Plus Capital Spending, SY2022-23",
    x = "Capital Share of Spending",
    y = NULL
  ) +
  theme_minimal()

See also

• The “CWIFT” article, if you want to compare labor costs across districts.
• The “CPI Adjustments” vignette for how flows are converted to constant dollars.
• The “Data Quality and Comparability” article for c11_spike_flag and the state-revenue adjustment.