WHO FUNDS UNIVERSITY RESEARCH? - DATA RELEASE
Release date: version 2026.10, October 2026
Online: https://ascendr.ai/resources/funds/

Full data behind the figures and tables of the report 'Who Funds University Research? Grants to US colleges
and universities in 13.9 million IRS Form 990 grant records, tax years 2020-2024, compared with federal and
foreign funding data' (George Koomullil).

REPORT
  who-funds-university-research.pdf (also at https://ascendr.ai/resources/funds/who-funds-university-research.pdf). The report's appendices list the leading research
  funders by field, the in-state funders of every state, funders active beyond R1, and peer-based leads
  for every R1 university; the tables here are their full versions.

SUGGESTED CITATION
  Koomullil, George (2026). Who Funds University Research? Data release, version 2026.10, October 2026.
  https://ascendr.ai/resources/funds/

LICENSE
  The project's compilation, documentation and derived columns: Creative Commons Attribution 4.0 International
  (CC BY 4.0, https://creativecommons.org/licenses/by/4.0/); cite the release as above. These licences cover only
  rights the project holds. Facts are not owned; grant text written by filers is distributed as public record from
  IRS returns and is not relicensed. Government sources (IRS, IPEDS,
  NCSES HERD, NIH, NSF, USAspending, Section 117) are public domain and are cited. Content derived from the 2025
  Carnegie Classification (column r1_2025 in file 04, the 'Newly R1 in 2025' rows and the r1_status column in
  file 08) is CC BY-NC-SA 4.0 with attribution to the Carnegie Classification of Institutions of Higher Education
  (ACE) and is not included in this release. The 2021 R1 flag is computed from IPEDS and is not affected.

SOURCES
  IRS Form 990 series e-file XML (Schedule I and Form 990-PF grant lists), IRS Exempt Organizations
  Business Master File, IPEDS HD2024, NCSES HERD Survey microdata, NIH RePORTER, NSF Award Search,
  U.S. Department of Education Section 117 records, and the 2021 and 2025 Carnegie Classifications.

CONVENTIONS
  - Amounts are US dollars, nominal. 990 amounts are grants reported as paid in the funder's tax year, 2020-2024,
    unless a file says otherwise; they are not comparable one-for-one with universities' fiscal years. Form 990-PF
    amounts are grants paid in cash or property; Schedule I amounts follow the filer's accounting method and can
    record a multi-year grant in the year it is made. Totals add the two, so every *_usd or amount_usd field holds
    reported grant amounts on mixed accounting bases, not a cash-flow or expenditure series.
  - 'Research' means an explicit research grant: the stated purpose names research (see the report's
    Methods and Appendix A). Grants whose research purpose is a building, construction, renovation, or
    facilities ('research capital') are not counted as research. External = excluding transfers from
    university foundations and affiliates, hospitals, athletic conferences, and government-funded
    intermediaries.
  - Institutions are IPEDS units (unitid); grants to a university's foundations and affiliates count for
    the university. Grants addressed only to a multi-campus system, with no evidence of the campus, are
    kept at system level: unitid starts with 'SYS' and the name ends '(campus not identified)'.
  - Recipient matching has been checked only on preliminary samples labelled by an AI model (about 89% of
    matches exact, 96% in the right institution family); no independent human check has been made, so
    institution-level figures are provisional rule-derived totals; check them against institutional gift records
    before quoting.
  - 'Research' is a classification of the grant's stated purpose text. Readings of the same texts by three
    language models and two human readers differ (see the report's validation appendix); the research
    figures are reported grant amounts whose text names research, not research expenditure and not a bound on it.
  - EINs and IPEDS unitids are identifiers, not numbers: read them as text (EINs are nine characters and can
    begin with 0; spreadsheet software may drop the leading zero when it treats the column as numeric).
  - FY2026 federal figures are provisional (data through 26 September 2026).
  - 'No grant' in the prospect-lead file means no grant observed in these data.
  - Federal, 990, and Section 117 measures are different quantities and must not be added together.
  - Files are UTF-8 CSV with a header row. True/False columns are booleans. Blank = not available.

FILES

data/01_research_funders.csv  (5,916 rows)
  Every private funder of explicit research grants to US universities, tax years 2020-2024. Report: Figure 5; Tables 8 and 17; Appendix D.
    funder_ein: IRS employer identification number of the funder
    funder_name: Funder name as filed (cleaned)
    funder_type: Funder type (see report Section 3.2)
    form: 990PF = private foundation return; 990 = public charity return
    funder_state: Funder's state (from its latest return)
    research_usd: Explicit research grants paid to US universities, 2020-2024, USD
    research_grants: Number of explicit research grants
    institutions: Number of institutions (IPEDS campuses) receiving them; grants resolved only to a multi-campus system are not counted as an institution
    r1_institutions_2021: Of which R1 (2021 Carnegie classification); r1_institutions_2021 + non_r1_institutions = institutions
    non_r1_institutions: Of which not R1 (2021)
    states_reached: Number of states of the receiving institutions
    median_grant_usd: Median research grant, USD
    top_field: HERD broad field with most of the funder's classified research dollars
    preselected_only: Yes = latest 990-PF states the foundation gives only to preselected organizations; Not stated = it does not say so; n/a = Form 990 filer (question not asked); Not available = latest return is a 990-PF whose application section could not be parsed
    first_year / last_year: First and last tax year with a research grant
    external_grants_to_universities_usd: All external grants (any purpose) to US universities, 2020-2024
    all_grants_paid_usd: All grants the funder paid to any recipient, 2020-2024
    assets_usd_latest: Fair market value (990-PF) or total assets (990), latest return
    website: Website reported on the return
  Note: Funders with a zero or negative net research total over 2020-2024 (refunds or reversals) are not listed: 2 funder(s), net USD -150,000. File 02 keeps their field rows, so the sum of file 02 is lower than the sum of file 01 by that amount; the report's national total uses the net figure.

data/02_funders_by_field.csv  (7,328 rows)
  Research grants by funder and estimated HERD broad field. Report: Figure 8, Table 9, Appendix B.
    field: Estimated HERD broad field (Unclassified = no field could be estimated)
    research_usd, research_grants, institutions, median_grant_usd: As in file 01, within the field
    funder_state, preselected_only: As in file 01

data/03_funders_by_state.csv  (14,950 rows)
  Research grants to the universities of each state, by funder. Report: Figure 16, Appendix C.
    institution_state: State of the receiving institutions
    funder_state: Funder's state
    funder_in_same_state: True when the funder is based in the same state
    research_usd, research_grants, institutions: Grants to that state's universities, 2020-2024

data/04_institutions.csv  (3,205 rows)
  Every US institution receiving external grants, 2020-2024. Report: Figures 7 and 11, Table 11.
    unitid: IPEDS institution identifier
    also_known_as: Common short names (e.g. UCLA, MIT, UNC), separated by semicolons; also in files 05-09
    control / carnegie_2021_group: IPEDS control; 2021 Carnegie group
    r1_2025: not in this file; not included in this release (see LICENSE)
    external_usd / external_funders: All external grants and number of external funders
    research_usd / research_funders / research_grants: Explicit research grants
    research_share_via_affiliates / external_share_via_affiliates: Share of dollars addressed to a foundation, medical school, or other affiliate
    herd_*_avg_fy2020_2024: HERD R&D expenditure, annual average FY2020-2024 (blank if not in HERD)
    research_per_year_over_herd_nonprofit: Annual research grants / HERD nonprofit-funded R&D: a ratio of reported grant amounts to nonprofit-funded R&D expenditure (different measures), not a coverage or recall rate

data/05_r1_exposure.csv  (146 rows)
  R1 universities (2021) by federal dependence and private research intensity. Report: Figure 19, Table 14.
    herd_*_avg_fy2023_2024: HERD R&D, FY2023-2024 average, USD
    federal_share_of_rd: Federal share of R&D
    research_grants_per_100_usd_rd: Annual explicit research grants per $100 of R&D
    research_funders: Number of research funders, 2020-2024
    quadrant: Position relative to the R1 medians of both measures

data/06_nih_awards_by_institution.csv  (611 rows)
  NIH extramural awards by institution, FY2020-FY2026 (FY2026 through 26 September 2026). Report: Figure 18, Table 13.
    nih_usd_fyYYYY / nih_awards_fyYYYY: NIH award dollars and number of awards
    nih_usd_fyYYYY_to_26sep / change_pct_fy2024_to_fy2026_same_date: NIH award dollars with award notice dates on or before 26 September of each fiscal year, and the change on that basis (used in the report)
    change_*_fy2024_to_fy2026: Change in NIH award dollars, full FY2024 against FY2026 to date (not like for like)
    private_research_grants_usd_per_year: Annual explicit research grants from 990 filers, 2020-2024

data/07_commitments_2024.csv  (3,430 rows)
  Grants approved for future payment (Form 990-PF, Part XIV line 3b) to US universities, tax year 2024: approvals made during the tax year and unpaid at the filer's year-end. The IRS instructions exclude approvals from earlier years that are still unpaid, so this is a cohort of new approvals, not the stock of outstanding commitments, and not available funds. Report: Figure 17, Table 12.
    committed_usd / committed_research_usd: Grants approved during tax year 2024 and unpaid at the filer's year-end (Form 990-PF line 3b; prior-year approvals not listed); of which explicit research
    grant_rows: Number of commitment lines

data/08_prospect_leads.csv  (5,913 rows)
  Peer-based prospect leads for every 2021 R1; the leads for institutions newly R1 in 2025 are not included in this release (see LICENSE). Report: Tables 18 and 16, Appendix E.
    r1_status: not in this file; the 'Newly R1 in 2025' rows and this column are not included in this release (see LICENSE)
    peers_funded_of_10: Number of the institution's ten peers the funder paid research grants to, 2020-2024
    research_paid_to_peers_usd: Research grants the funder paid to those peers
    definition: A lead funded at least three peers and has no grant of any kind (paid or committed, 2020-2025) to the institution observed in these data (missing grant lists or unresolved recipient names can hide a relationship); single-state funders are leads only in their state
  Note: Institutions in scope with no lead under the definition are not rows in this file: University of Rhode Island (217484). A lead is a historical relationship observed at peers, for manual verification; it does not establish current eligibility or an open application route (see preselected_only), and the evidence window is tax years 2020-2024 for peers' grants and 2020-2025 for the absence of a grant to the institution.

data/09_section117_by_institution.csv  (14,841 rows)
  Foreign gifts and contracts reported under Section 117, by institution, year, and country, 2020-2025. Report: Figure 20, Table 15.
    gift_or_contract: Transaction type as reported
    foreign_government: True when the source is a foreign government
    amount_usd / records: Reported amount and number of records
  Note: File 09 lists only records whose institution was matched to an IPEDS unit; 3 record(s) worth USD 834,742.00 (raw amount) could not be matched and appear in file 10 (by country) but not here. Amounts in files 09 and 10 are rounded to whole dollars per row, so their totals differ from this raw figure and from each other by rounding residuals of a few dollars.

data/10_section117_by_country.csv  (1,641 rows)
  Section 117 foreign gifts and contracts by country and year, 2020-2025. Report: Figure 20.
    n_institutions: Number of reporting institutions

FIELD CROSSWALK
  field_crosswalk.csv: the keyword rules (Python regular expressions, applied to the upper-cased purpose
  text) that assign a HERD broad field in files 02 and 04. A text is assigned the field whose pattern
  matches most often; ties go to the earlier rule (rule_order). Nine of HERD's ten broad fields have a
  rule; 'Other sciences (nec)' (HERD's sciences, n.e.c., about 1% of R&D) has none and is listed with
  assigned = False. Texts matching no rule are 'Unclassified' and take the funder's dominant field
  where the report says so. The field label is a keyword estimate and has not been validated. The stems LABOR,
  ELECTION, DESIGN, VISION, AGING, CARDI(O), DENTAL, PAIN, LIVER and AIDS are bounded to whole words (an unbounded
  LABOR would match COLLABORATION).

Also included: who-funds-university-research-data.zip (all files above, this README and the crosswalk).
