Nonprofits in the Milwaukee metro, 1989–2026: counts, focus areas, and revenue

The ask

For the Milwaukee MSA — most recent snapshot and history back to 1989: how many nonprofits overall and by county; how many by focus area; revenue by focus area; revenue by source; and revenue by focus area × source.

(Requester identity and the verbatim ask are in the gitignored _private.md.)

Geography — the Milwaukee MSA

“Milwaukee MSA” resolves to OMB CBSA 33340 “Milwaukee-Waukesha, WI” — four counties: Milwaukee, Waukesha, Ozaukee, Washington — held fixed (2023 OMB vintage) across all years so the series is comparable. Geography is composed by the canonical crosswalk join only, never re-derived:

BMF (geo_state_abbr, geo_county)
  └─ join county-fips crosswalk  → geo_county_fips
       └─ join cbsa crosswalk    → cbsa_code, filter == "33340"

Data & method

• Counts & focus area → geocoded BMF master (nccs_read): one row per EIN (3.67M), cumulative across vintages, each carrying first_year_in_bmf / last_year_in_bmf (span 1989–2026), geography, and NTEE. Defunct orgs are retained. “Registered in year Y” ⟺ first_year_in_bmf ≤ Y ≤ last_year_in_bmf.
• Revenue → CORE 990 merged tier (nccs_read_core). CORE carries EIN + financials only — geography/NTEE come from joining CORE → master on ein (consumer-composed, ADR 0016). Public charities (990/990-EZ → 990combined) and private foundations (990pf) are read separately and reported separately.
• Focus area = nteev2_subsector (Arts, Education, Environment, Health, Human Services, Public/Societal Benefit, Religion, …). Chosen over the legacy major group because it classifies far more revenue — see caveat 3.
• Revenue sources (990combined): contributions & grants (total_contributions), program services (program_service_revenue), fundraising events — net (net_income_fundraising_events), investment (investment_income), other (other_revenue_total_11e).
• Dollars: both nominal and CPI-adjusted (real 2024 $, BLS CPI-U annual avg).

Coverage caveats (validated against the data — read before citing)

1. Two populations. Counts = the registered universe (BMF, all org types). Revenue = filers only — Form 990 + 990-EZ (public charities, ~2,800/yr recently) and Form 990-PF (private foundations, ~1,600/yr recently), reported separately. Orgs below the filing threshold (and Form 990-N postcard filers) report no financials.
2. Historical geocoding decays with age. Share of then-active WI orgs with a resolvable county: 1989 40% · 2005 61% · 2015 83% · 2020 93% · 2024+ 100%. Recent snapshot is exact; pre-2010 by-county/focus counts undercount — read early years as trend/lower-bound.
3. ~⅓ of revenue is unclassified. Even with nteev2_subsector, “UNU” (unclassified) holds 33% of 2023 Form 990/990-EZ revenue (and ~36% of the current count). The legacy major group was worse (63% unclassified). UNU is reported as its own row, not dropped.
4. Count series is a step function at vintage years. Vintages are sparse pre-1995 (only 1989, then 1995), so 1990–1994 repeat the 1989 value; resolution improves over time.
5. 2024 is partial (filings still arriving) — the latest complete revenue year is 2023. Snapshot tables use 2023.
6. Revenue-source fields drift over time (dues folded into program revenue ~2008; “events” is net of direct expenses); classification is latest-observed per EIN.
7. Legacy-year revenue reflects upstream completeness. Pre-2012 filer counts are depressed by the geocoding gap (caveat 2) plus thinner legacy ingest; several years dip (e.g. 1999, 2009, 2011; Form 990-PF 2016–2018 is sparse). Read pre-2012 dollars as directional, not exact.

bmf <- read_bmf_master(
  req, state = "WI",
  columns = c("ein", "geo_state_abbr", "geo_county", "nteev2_subsector",
              "first_year_in_bmf", "last_year_in_bmf"),
  cache = FALSE,   # read S3 directly with predicate pushdown (no 400MB cache)
  note = "WI universe, all vintages 1989-2026"
)
xwalk_county <- read_crosswalk(req, "county-fips")
xwalk_cbsa   <- read_crosswalk(req, "cbsa")

msa <- bmf |>
  left_join(xwalk_county, by = c("geo_state_abbr", "geo_county" = "geo_county_raw")) |>
  left_join(xwalk_cbsa,   by = c("geo_county_fips" = "county_fips")) |>
  filter(cbsa_code == "33340") |>
  mutate(focus = ifelse(is.na(nteev2_subsector) | nteev2_subsector == "", "UNU", nteev2_subsector))

active <- function(df, y) filter(df, first_year_in_bmf <= y, last_year_in_bmf >= y)

Findings

Q1 — How many nonprofits, overall and by county

Registered nonprofits in the Milwaukee MSA (geocoded; selected years):

Current (2026) by county: Milwaukee 6,918 · Waukesha 2,717 · Ozaukee 695 · Washington 694 (total 11,024). Early-year figures undercount — caveat 2.

Q2 — By focus area (current, 2026)

Q3 — Revenue by focus area (2023, Form 990/990-EZ, $M)

Health systems dominate metro nonprofit revenue.

Q4 — Revenue by source (Form 990/990-EZ, $M)

Total 990/990-EZ revenue: nominal $1.5B (1989) → $27.2B (2023); in real 2024 dollars $3.8B → $28.0B (~7×), with a 2021 dip to $19.2B nominal. Program service revenue is ~80% of the total — the hospital/health footprint. (Full 35-row series in data/revenue_by_source_by_year_1989-2023.csv.)

Q5 — Revenue by focus area × source (2023, Form 990/990-EZ, $M)

Human services lean on contributions; health/education on program revenue.

Private foundations (Form 990-PF), reported separately ($M)

Form 990-PF revenue grew from ~$0.1B (1989) to ~$1.2B (2023) nominal. (2016–2018 is sparse upstream — merged-tier 990pf backfill gap; full series in data/q_pf_total_revenue_by_year.csv.)

# Full reproducible computation lives in data/_compute.R (writes the CSVs the
# tables above are drawn from). It reads CORE 990combined (990/990-EZ) + 990pf
# (990-PF) for 1989:2023 via the merged tier (per-year reads to absorb legacy
# schema drift), joins the MSA universe on EIN, and aggregates by year / focus /
# source with CPI deflation.

Takeaways

• The four-county Milwaukee metro has ~11,000 registered nonprofits today, roughly 6× the geocodable 1989 count (early counts undercount).
• Human services, public/societal benefit, education, and religion are the most numerous; health systems dominate revenue (~$10B of ~$27B in 2023).
• Form 990/990-EZ (public-charity) revenue grew from ~$1.5B (1989) to ~$27.2B 2023) nominal — ~$3.8B → ~$28.0B in real 2024 dollars (7×). Program service revenue is the overwhelming source (~80%); contributions are second ($3.4B in 2023).
• Form 990-PF (private foundations) add ~$1.2B (2023), reported separately.

Reproducibility: reads the vintages pinned in _pins.csv. See ADR 0024 / 0025 in nccs-contracts.