USDA over-counts
foundation support to
rural America by 78%.

USDA's $3.5B "rural philanthropy" number
measures the wrong thing.

USDA Rural Development's March 2026 report Rural America's Philanthropic Sector concludes that rural-based grantmaking organizations issued $3.5 billion per year in grants from 2014 to 2021 — about 3% of the $116B annual U.S. total. The framing throughout the report treats that $3.5B figure as a measure of philanthropic support for rural America. It is not. It is a measure of dollars disbursed by foundations whose headquarters sit in non-metro counties, regardless of where those dollars ultimately land.

Our destination-based accounting of the same underlying IRS data tells a different story. We re-parsed all 1.72 million individual grants in the 2023 IRS 990-PF e-file, joined each recipient ZIP to its USDA Rural-Urban Commuting Area classification, and measured where dollars actually land. The result: 44% of grant dollars disbursed by rural-headquartered foundations leave rural America to fund metropolitan recipients (mostly large universities, national arts organizations, and denominational headquarters). Apply that leakage rate to USDA's $3.5B figure and the true rural-to-rural foundation flow is closer to $1.96 billion per year. USDA's number is 178% of the destination-based reality — an overcount of 78%.

This report makes the case in three parts. Section I documents what USDA reports and identifies the three methodological choices that drive the overcount — source-based counting rather than destination-based, county-MSA classification rather than ZIP-level RUCA, and the absence of any recipient-side analysis in the underlying IRS-data pipeline USDA used. Section II presents the destination-based picture itself: where the $125.5 billion in 2023 U.S. foundation grant dollars actually landed, by ZIP-level RUCA tier, and what that view shows that USDA's view cannot. Methodology and replication detail follow.

The underlying critique is not that USDA's analysts made an error. The 2014–2021 grant-source aggregation is internally consistent and answers a legitimate question (where are grantmaking organizations headquartered?). The critique is that USDA presents that source-based number as a measure of rural philanthropic support, and that policymakers and funders reading USDA's report are likely to conclude that rural America already receives $3.5 billion per year in foundation grants when in fact a substantial share of that money never reaches a rural recipient.

What USDA reports, and why
the $3.5B figure overcounts.

USDA Rural Development's Rural America's Philanthropic Sector (March 2026, authored by Nicholas Chiumenti et al., Innovation Center / Data Analytics Division) is a careful internal analysis of the IRS Form-990 and Form-990PF e-files for tax years 2014–2021. The report's central empirical finding is reproduced below. The critique that follows is not about the underlying counts — which are correctly computed from the source data — but about the methodological framework that turns those counts into a headline claim about philanthropic support for rural America.

What USDA's report finds

USDA identifies 75,242 active grantmaking organizations in the U.S. that issued an average of $116 billion in grants per year between 2014 and 2021. Of those organizations, USDA reports:

• 8% (~6,300 organizations) are headquartered in rural communities, defined as counties outside a Metropolitan Statistical Area.
• $3.5 billion in grant dollars per year were issued by these rural-headquartered organizations — about 3% of the $116B annual total.
• One-third of rural counties have no locally-based grantmaking organization at all.
• Half of all rural-based grantmaking organizations are in just 13 states.
• Rural per-capita grantmaking ranges from $56/resident (rural South) to $122/resident (rural Northeast); urban per-capita ranges from $271 to $533.

USDA's report acknowledges, on page 18, that this is a source-based view: "the data used in this report only includes the total amount of grants made by the organization… does not include data on the specific purpose of the grants made by these organizations, the location of the grant recipient, or the area being served by the grant." The report goes on to note that "a rural foundation may issue a grant to an urban-based recipient to fund services that in turn will be provided in a rural community."

That caveat is correct as far as it goes. What the report does not do — and what produces the overcount — is quantify how large the rural-to-metro outflow actually is, or revise the headline numbers in light of it.

The three methodological choices that produce the overcount

1. Source-based, not destination-based. USDA's headline "$3.5B in rural philanthropy" is the sum of grant dollars disbursed by rural-headquartered foundations — not the sum of grant dollars landing in rural recipients. The two are very different. Foundations headquartered in rural counties regularly send grants to recipients outside rural America: alumni gifts to flagship metropolitan universities, donations to national arts organizations, contributions to denominational headquarters, regrants to large urban intermediaries.

To quantify this, we parsed the 1.72 million individual grants in the 2023 IRS 990-PF e-file, extracted each recipient address, and joined to ZIP-level RUCA. Of the $1.23 billion disbursed by Small Town and Rural-headquartered private foundations combined, $537 million (43.7%) flowed back to Big City ZIPs. Another $18M went to foreign and unclassified destinations. Only $675M (54.9%) stayed in non-Big-City destinations — and only a portion of that stayed in the originating tier. Applying the 44% leakage rate to USDA's $3.5B figure yields a true rural-to-rural flow of approximately $1.96 billion per year, not $3.5 billion.

2. County-level MSA classification, not ZIP-level RUCA. USDA defines a county as "rural" if it sits outside a Metropolitan Statistical Area. Roughly 46 million Americans live in those non-MSA counties. But non-MSA counties are not uniformly rural at the ZIP level — they contain regional centers (cities of 10,000–49,000 population that serve as the hub for surrounding rural ZIPs). Marshalltown IA, Pierre SD, Eau Claire WI, and dozens of similar towns are classified as "rural" under USDA's framework even though they are operationally distinct from the truly rural ZIPs (RUCA 10) that surround them.

These regional centers receive substantially more foundation grants per capita than truly rural ZIPs — they're where the hospital systems, community colleges, and community foundations tend to sit. Pooling them with the surrounding RUCA 7–10 ZIPs pulls the rural per-capita average up. The denominator is also wrong in the other direction: USDA's "urban" universe (~285M) includes the truly rural ZIPs that happen to sit inside metro counties — Polk County (IA), Loudoun County (VA), Yolo County (CA), and many others have RUCA 9–10 ZIPs that USDA classifies as urban.

3. No recipient-side data in the underlying pipeline. The IRS Form 990 and 990-PF aggregate financial extracts USDA used contain organization-level totals (total grants paid by the foundation) but not the individual grant records. The Form 990-PF schema contains a per-grant GrantOrContributionPdDurYrGrp element with recipient address fields, but it requires re-parsing the raw XML rather than relying on the IRS pre-aggregated extracts. USDA chose the simpler aggregate-extract approach. The trade-off is that the resulting dataset cannot answer destination-side questions at all, and the report is structurally forced into a source-based framing.

Why the methodological choice matters: rural quality-of-life indicators

The disagreement between source-based and destination-based accounting is not academic. It determines which communities the philanthropic sector recognizes as underserved, and at what scale. The communities at stake are demonstrably struggling on a range of measurable quality-of-life indicators. U.S. Census Bureau American Community Survey data and the 2025 edition of the County Health Rankings & Roadmaps (a joint project of the Robert Wood Johnson Foundation and the University of Wisconsin Population Health Institute, drawing on federal data from CDC NCHS, CMS, and the Census Bureau) show consistent and substantial rural-urban gaps. Population-weighted averages across the 1,958 non-metropolitan counties (RUCC 4–9, home to ~46 million Americans) compared to the 1,186 metropolitan counties (RUCC 1–3, home to ~285 million Americans):

• Life expectancy: 74.6 years in non-metro counties vs 77.7 in metropolitan counties — a 3.1-year deficit. (Source: CDC NCHS via County Health Rankings 2025.)
• Poverty rate: 15.4% in non-metro vs 12.1% in metropolitan — 3.3 percentage points higher. (Source: U.S. Census ACS 5-year estimates.)
• Premature death: 10,826 years of potential life lost per 100,000 residents in non-metro vs 8,069 in metropolitan — 34% higher. (Source: CDC NCHS via County Health Rankings 2025.)

These indicators are the empirical signature of what Princeton economists Anne Case and Angus Deaton termed "deaths of despair" — the rising mortality among working-age Americans from alcohol, drug overdose, and suicide that began in the late 1990s and accelerated through the 2010s. The non-metropolitan suicide rate is 19.8 per 100,000 vs 13.3 in metropolitan counties — 49% higher. Drug-overdose deaths are roughly comparable in raw rate (31.4 metro vs 32.2 non-metro per 100,000), but Case and Deaton's broader argument is that the trajectory in non-metro communities has been steeper, has affected younger cohorts more severely, and has been more concentrated in places with declining labor-market opportunity.

A substantial academic literature has documented that the same county-level health indicators predict the shift in rural counties' two-party presidential margins from 2008 through 2024 — net of demographic and economic controls. Shannon Monnat's work on the 2016 election (Carsey School of Public Policy / Pennsylvania State University) and related work by Anne Case, Angus Deaton, and colleagues at Princeton's Center for Health and Wellbeing have shown that drug-overdose, suicide, and alcohol-related mortality rates were among the strongest county-level predictors of the rural Republican margin gain.

Our own analysis of county-level results bears the pattern out. Among the 1,929 non-metropolitan counties for which we have matched health and election data, the Pearson correlation between county-level poverty rate and the 2020→2024 two-party margin shift toward the Republican candidate is +0.40; between premature death rate and the same margin shift, +0.29; between life expectancy and the same shift, −0.28. Counties with worse health and economic outcomes shifted more decisively toward the candidate who promised to address them. The political interpretation of that correlation is contested; the correlation itself is not. What is largely uncontested is that the underlying conditions — the lower life expectancy, the higher poverty rate, the higher premature death rate — are real, measurable, and require resources to address.

Foundation philanthropic capital is one source of those resources. The question of how much actually flows into these communities is therefore not a methodological abstraction. It is the difference between recognizing the scale of the gap and underestimating it.

RUCC vs RUCA: the two federal classifications of rural America

USDA's report defines "rural" using one federal classification system; this report uses a different one. Both are official; both are maintained by federal agencies; they measure different things and produce different counts. Understanding the distinction is essential to reading either report carefully.

The Rural-Urban Continuum Codes (RUCC), maintained by the USDA Economic Research Service, classify each U.S. county on a 1–9 scale by metropolitan-area population size and adjacency. Codes 1–3 designate metropolitan counties by metro-area size (1 million or more residents, 250,000 to 1 million, and fewer than 250,000, respectively). Codes 4–9 designate non-metropolitan counties by the size of their largest urban population and by whether the county is adjacent to a metro area: codes 4 and 5 cover counties with an urban population of 20,000 or more (adjacent and non-adjacent to metro, respectively); codes 6 and 7 cover counties with an urban population of 2,500 to 19,999; codes 8 and 9 cover completely rural counties or those with an urban population of fewer than 2,500. RUCC operates at the county level only. A county is classified entirely as metro or non-metro based on its largest urban area and Office of Management and Budget metropolitan-statistical-area boundaries.

The Rural-Urban Commuting Area codes (RUCA), also maintained by the USDA Economic Research Service in collaboration with the federal Health Resources & Services Administration's Federal Office of Rural Health Policy, classify each U.S. census tract (and, in a derived version, each U.S. ZIP code) on a 1–10 scale combining settlement size with daily commuting flows from the U.S. Census Bureau's American Community Survey. Codes 1–3 designate metropolitan areas (Census-Defined Urbanized Areas of 50,000 or more, and tracts with primary commuting flows toward those Urbanized Areas). Codes 4–6 designate micropolitan areas (Urban Clusters of 10,000 to 49,999 population and tracts with commuting flows toward them). Codes 7–9 designate small-town areas (Urban Clusters of 2,500 to 9,999 population and their commuting flows). Code 10 designates pure rural areas — those with no urbanized area or urban cluster within commuting distance.

The practical difference for this analysis: RUCC describes the whole county and has no sub-county granularity. It cannot distinguish a rural ZIP inside a metro county (Polk County IA has several; Loudoun County VA has many) from the metro core of the same county. RUCA describes the ZIP and can. For analyzing where foundation grant dollars land — the destination side of the philanthropic flow — RUCA's ZIP-level granularity is the appropriate tool, because the recipient address in the IRS Form 990-PF grant record is a ZIP code, not a county. USDA's choice of county-level RUCC effectively rounds every recipient ZIP up or down to its county's metro classification, losing the sub-county rural detail that RUCA preserves.

The defensible part of USDA's report

USDA's analysis answers a legitimate and important question: where in the country are grantmaking organizations physically headquartered? The finding that 8% of grantmakers sit in rural counties, that one-third of rural counties have no local grantmaker, and that half of all rural-based grantmakers cluster in 13 states is genuinely useful for understanding the geography of philanthropic capacity. The Federal Reserve Bank of Philadelphia study USDA cites — finding that 71–91% of community-foundation grants stay within their immediate metro or county — suggests that organizational capacity is correlated with destination, even if the correlation is imperfect.

The disagreement is not with the underlying analysis. It is with the framing. USDA's headline numbers are presented — in the report's introduction, conclusion, and derivative communications — as a measure of philanthropic support for rural communities. They are not. They are a measure of rural philanthropic capacity, which is a much narrower claim. Section II below presents the destination-based view USDA's framework cannot.

The destination-based picture,
by RUCA tier.

Where does U.S. foundation grant money actually land? To answer that question we re-parsed every grant in the IRS 990-PF e-file for tax year 2023 — 126,982 private foundations and 1.72 million individual grants — and joined each recipient ZIP to its USDA Rural-Urban Commuting Area classification. The result is the four-tier picture USDA's methodology cannot produce: Big City ZIPs (RUCA 1–3) received 87% of all U.S. foundation grant dollars in tax year 2023. The other three tiers combined — Small City, Small Town, and Rural — received 4.1%.

The U.S. population baseline, by RUCA tier

Before reading the grant-flow distribution below, it is worth fixing the population baseline. Roughly 80% of Americans live in Big City ZIPs (RUCA 1–3, Census-Defined Urbanized Areas). The remaining 20% — about 65 million Americans — live in Small City, Small Town, or Rural ZIPs. This is the comparison every grant-flow figure in this section should be read against.

Why Rural ZIPs outperform Small Town ZIPs per capita

The per-capita chart above shows Rural ZIPs (RUCA 10) receiving $77 per resident in foundation grants while Small Town ZIPs (RUCA 7–9) receive only $54. That looks counter-intuitive — you'd expect a monotonic decline from Big City through Small City and Small Town down to Rural. It isn't an error. Three distinct mechanisms drive it, in order of contribution:

1. The denominator effect. Rural ZIPs hold roughly 12.4 million U.S. residents; Small Town ZIPs hold 23.2 million — almost twice as many. Equivalent dollar amounts spread across a smaller population produce a higher per-capita figure. Even if both tiers received the same total grants, Rural's per-capita would lead by 1.9× on the math alone.

2. Anchor institutions concentrate in deeply rural ZIPs. Small private liberal-arts colleges, cultural destinations, conservation organizations, and religious orders tend to locate in RUCA 10 settings rather than RUCA 7–9 Small Town settings. The top Rural-tier recipients in tax year 2023 include Alice Lloyd College (KY), Sterling College (VT), Colby Sawyer College (NH), Young Harris College (GA), Sweet Briar College (VA), Bigelow Laboratory for Ocean Sciences (ME), Fenimore Art Museum in Cooperstown NY, Grand Teton National Park Foundation (WY), Mount Desert Island Hospital (ME), and Marians of the Immaculate Conception (MA). Each of these is a national-donor magnet sitting in a sub-2,500-population RUCA 10 ZIP — lifting Rural's per-capita by several dollars per institution.

3. Data-quality artifacts — particularly pharma patient-assistance foundations. The #1 Rural-tier "recipient" in tax year 2023 is a placeholder entry labeled "Donations US" at ZIP 56623 (Baudette, MN), valued at $76.92 million — 8.1% of the entire Rural tier total. The source is the ANI Rare Disease Patient Assistance Program, the charitable arm of ANI Pharmaceuticals (whose corporate headquarters has been in Baudette for decades). The $76.92M does not represent foundation grants flowing into Baudette: it represents pharma product-cost donations to patients across the country, accounted for as a single aggregate "grant" from a Rural-ZIP foundation to a placeholder recipient in the same ZIP. Several other placeholder entries in the Rural list ("See Schedule Attached", "See Schedule A", "See Statement Regarding Dissolution") together account for another ~$40M. Combined, data-quality artifacts inflate the Rural tier total by roughly 12 percent.

Net effect. Stripping the placeholder and pharma-patient-assistance artifacts brings Rural per-capita down to approximately $67 per resident — still meaningfully above Small Town's $54, but the gap narrows from 1.43× to about 1.24×. The Small-Town-is-worst-served finding is robust to data cleanup; the Rural-beats-Small-Town finding is real but partly an artifact of small denominators, a few well-funded anchor institutions, and a pharma accounting convention that lands in one Minnesota ZIP.

Ten facts from the IRS 990-PF data

1. Of every $100 in U.S. foundation grants, Big City ZIPs get $87. Total grant dollars traced in tax year 2023: $125.5B. Big City (RUCA 1–3) received $109.2B (87.0%). Small City (RUCA 4–6) received $2.94B (2.34%). Small Town (RUCA 7–9) received $1.26B (1.00%). Rural (RUCA 10) received $951M (0.76%). The remainder went to foreign recipients ($6.99B, 5.57%) or to recipients whose ZIPs could not be classified ($4.15B, 3.30%).

2. Of grant dollars with a confirmed U.S. destination, 95.5% land in Big City ZIPs. Filtering out the foreign and unknown destinations, $114.4B reached a confirmed U.S. recipient. Big City: $109.2B (95.5%). Small City: $2.94B (2.6%). Small Town: $1.26B (1.1%). Rural: $951M (0.83%). The three non-Big-City tiers combined hold roughly 20% of the U.S. population and receive 4.5% of U.S. foundation grant dollars by RUCA-level analysis.

3. The per-capita gap is 7.6× at its worst point. Big City: $411 per resident per year. Small City: $100. Rural: $77. Small Town: $54. The Big City vs Small Town ratio of 7.6× is the sharpest per-capita disparity. A Small Town resident receives roughly 13 cents on the Big City dollar in U.S. foundation grant funding.

4. Big City foundations pay 97% of U.S. foundation grant dollars. 113,761 Big City-headquartered private foundations disbursed $122.3B in tax year 2023. The other three tiers combined — Small City (6,765 foundations), Small Town (3,623), Rural (2,448) — disbursed $2.91B (2.4%). Big City foundations outweigh the rest of the country by ~42× in annual grant volume.

5. Big City foundations send only 2.9% of their grant dollars outside Big City ZIPs. Of the $122.3B distributed by Big City foundations, $107.8B (88.1%) stayed in Big City ZIPs. Small City received $2.07B (1.7%), Small Town received $883M (0.7%), Rural received $638M (0.5%) — together $3.59B or 2.9% of Big City foundation outflows. Foreign recipients received $6.82B (5.6%); unknown destinations $4.13B (3.4%).

6. Rural ZIP organizations depend on Big City money for 67% of what they receive. Of the $951M reaching Rural ZIPs in tax year 2023: $638M (67%) came from Big City foundations; $239M (25%) came from Rural foundations themselves; $73M (8%) came from Small City and Small Town foundations combined. Rural civic infrastructure is structurally dependent on metropolitan philanthropic capital it does not control.

7. Small Town ZIPs are the most under-served tier per capita. Despite holding roughly 7% of the U.S. population (23.2M residents), Small Town ZIPs (RUCA 7–9) received 1.0% of all U.S. foundation grant dollars. Per capita: $54. Pure Rural ZIPs (RUCA 10) at $77 per resident benefit from a handful of land-trust, conservation, and rural-health grants that bypass the Small Town tier entirely.

8. Rural and Small Town foundations send a significant share back to Big City ZIPs. Of the $1.23B disbursed by Small Town + Rural-headquartered foundations combined, $537M (44%) flowed back into Big City ZIPs — mostly through large metropolitan-headquartered recipients (universities, national arts organizations, large national charities). Only $574M (47%) stayed in Small City, Small Town, or Rural ZIPs combined.

9. The conventional "metro / non-metro" framing understates the gap. The county-level RUCC classification puts 96% of foundation grant dollars in "metro" counties. The ZIP-level RUCA classification puts 95.5% in Big City ZIPs alone — before counting Small City ZIPs that the RUCC metro bin treats as metro but that are operationally distinct from major-metro Big City ZIPs. The four-tier RUCA view exposes a more severe disinvestment in non-Big-City America than the standard two-tier RUCC view conveys.

10. Foreign destinations receive more U.S. foundation grant dollars than all Rural ZIPs combined. Foreign recipients: $6.99B (5.57% of all grants). Rural ZIPs: $951M (0.76%). The U.S. private foundation sector sends 7.4× more grant dollars outside the U.S. than to all Rural ZIPs in the U.S. combined. Foreign destinations receive 5.5× more than all Small Town + Rural ZIPs combined ($2.21B).

Headline numbers

Structural challenges that perpetuate the divide

Foundations frequently respond to rural-funding-gap critiques by noting that they fund what they receive proposals for, and that rural organizations submit fewer proposals. This is true and circular. The professionalized grant-writing infrastructure that converts community needs into funded work is itself a Big-City civic asset. Most non-Big-City ZIPs do not have a paid grant writer, a community foundation with serious staff, a development director at a local hospital, or an advocacy nonprofit with full-time fundraising capacity. The proposal pipeline that funders depend on has been built in Big City ZIPs, by Big City organizations, around Big City grantmakers' calendars and templates. The non-Big-City civic infrastructure documented in the Iowa companion report — Farm Bureaus, volunteer fire departments, PTAs, electric cooperatives, volunteer-run nonprofits — is overwhelmingly volunteer, lay-led, and not staffed for foundation-style proposal work.

The result is a flywheel of disinvestment: non-Big-City civic organizations cannot afford to professionalize because they cannot raise foundation money, and they cannot raise foundation money because they have not professionalized. The structural-recommendation work for closing the gap — non-Big-City philanthropic-intermediary capacity, community-foundation capitalization in Small City and Small Town ZIPs, place-based national-foundation program officers physically embedded outside Big City ZIPs, and proposal-pipeline investment scaled for non-Big-City budgets — is what the gap actually requires.

How we did this.

Data sources

• IRS Form 990-PF e-file, tax year 2023. Pulled from the IRS Tax-Exempt Organization Search bulk e-file pipeline. 126,982 private-foundation filings parsed with the Rural Desk 990-PF parser (polling_and_data_research/fetch_990pf_grant_flows_ruca.py). Filer-scoped extraction (the actual foundation, not the tax preparer). Self-dealing flags and grantor-sibling re-grants excluded where identifiable.
• USDA Economic Research Service RUCA2010 ZIP Code file. ZIP-level Rural-Urban Commuting Area codes for all U.S. ZIPs (~41,000 total). Each recipient ZIP in the 990-PF file is joined directly to a primary RUCA code; the code maps to one of four tiers (Big City RUCA 1–3, Small City RUCA 4–6, Small Town RUCA 7–9, Rural RUCA 10).
• 2020 U.S. Census population by ZIP. Used to compute the per-capita grant-receipt figures in Section I. Populations are aggregated to RUCA tier and then divided into the per-tier grant total.

Geographic classification — ZIP-level RUCA

Each grant in the 990-PF e-file carries a recipient address. For U.S. recipients we extract the ZIP and join directly to USDA RUCA2010. This is a meaningful improvement over the conventional ZIP → County FIPS → RUCC chain used in most national rural funding analyses, which collapses ZIP-level rural texture into the metro/non-metro binary of its parent county. Rural ZIPs inside metro counties (Polk County, IA has several; Loudoun County, VA has more) are visible under the RUCA framework and hidden under the RUCC framework.

Tiers:

• Big City — primary RUCA 1–3 (Census-Defined Urbanized Areas; ~266M residents)
• Small City — primary RUCA 4–6 (Large Urban Clusters, 10K–49,999 population; ~30M residents)
• Small Town — primary RUCA 7–9 (Small Urban Clusters, 2,500–9,999 population; ~23M residents)
• Rural — primary RUCA 10 (areas outside any urbanized area or cluster; ~12M residents)

Grant amount extraction

Each individual grant record (GrantOrContributionPdDurYrGrp in the 990-PF XML schema) is parsed for amount, recipient name, and recipient address. Grants with malformed amounts (non-numeric, blank, or null) are skipped silently. Grants with no recipient address are classified as "unknown destination" rather than discarded, so the totals are conservative: the $4.15B in "unknown destination" represents real grant dollars that flowed somewhere, just to a recipient whose ZIP we could not classify.

Per-capita calculation

Per-capita figures divide the total grant dollars landing in each RUCA tier by the 2020 Census population resident in that tier. The denominator is conservative: U.S. Census ZIP-level population aggregated to primary RUCA code, before any adjustment for cross-tier commuting or partial-overlap ZCTA boundaries.

Limitations and caveats

• ~3% of grant dollars have no parseable destination ZIP. These are tracked separately as "unknown destination" rather than redistributed, so per-tier totals are floor estimates.
• ~6% of grant dollars go to foreign recipients. The U.S. foundation sector's foreign grant-making is real but outside the rural-domestic scope of this report.
• ZIP → RUCA join is an approximation for service area. A statewide rural-development nonprofit headquartered at a Big City ZIP serves rural communities operationally; this report counts it as a Big City recipient because the grant landed in a Big City ZIP. The geographic pattern documented here describes where the institutional infrastructure sits, not where the service is ultimately delivered.
• Filer-scoped extraction. A previous version of this pipeline had a bug that attributed foundation grants to the tax preparer's address rather than the foundation's address. The current pipeline scopes all source-side extraction to the <Filer> element.
• Tax year 2023 only. The 990-PF e-file aggregation is annual; this report uses tax year 2023 (filings received in calendar 2024). Multi-year trend analysis is out of scope here.

Replication

The full data pipeline — 990-PF e-file download, XML parse, ZIP → RUCA join, per-tier aggregation, and JSON export — runs in a single Python script (fetch_990pf_grant_flows_ruca.py) against publicly available source files. Outputs include a 4×6 flow matrix CSV, a per-foundation atlas JSON, and the headline aggregates JSON consumed by the charts on this page. The pipeline runs in about 25 minutes on a standard developer laptop against the full ~4.5 GB cached 990-PF XML archive.