Federal research funding is highly concentrated across universities (Gini ~0.82; top 10 get ~20%, top 100 get ~80%) — and concentration among the very top has not eased over the past decade

Summary

Federal research funding is very concentrated across U.S. universities: in FY2024, the top 10 institutions (out of 674 receiving any federal R&D money) captured about 21% of all federal R&D dollars, the top 20 captured 35%, and the top 100 captured 81%. The Gini coefficient across all 674 institutions is 0.82 — an 'extreme' level of concentration on the standard 0 (equal) to 1 (all-to-one) scale. Over the past decade this concentration has NOT worsened by the simple top-10-share measure — NCSES's own trend report shows the top 10 recipients' share of federal science-and-engineering support actually eased slightly, from 22.7% in FY2014 to 20.1% in FY2023 — but the gap between the single largest recipient (Johns Hopkins University) and the #2 recipient widened sharply over the same period, from $764 million to nearly $1.2 billion, indicating the very top of the distribution pulled further ahead even as the broader top-10 group's share shrank a bit.

How concentrated is it today (FY2024, warehouse computation)

Using NCSES's Higher Education R&D (HERD) Survey microdata (research.nsf_herd_by_institution, the only year currently loaded in this warehouse is FY2024), 674 institutions reported nonzero federal R&D expenditure, totaling $64.6 billion:

• Top 10 institutions: $13.47B, or 20.8% of the total
• Top 20 institutions: $22.36B, or 34.6%
• Top 100 institutions: $52.49B, or 81.2%
• Remaining ~574 institutions: 18.8%

The leading recipient, Johns Hopkins University (whose federal total includes the Applied Physics Laboratory, a University Affiliated Research Center), received $3.62 billion in FY2024 — more than double the #2 recipient, Georgia Tech, at $1.21 billion. The Gini coefficient across all 674 institutions is 0.82, in the 'extreme' concentration band; for comparison, this is a higher inequality reading than most national income-distribution Gini figures (typically 0.35-0.5), reflecting that federal research money is not spread anywhere close to evenly — hundreds of institutions receive only trace amounts while a small number of large research universities and federally funded R&D centers dominate.

The Herfindahl-Hirschman Index (HHI) computed the same way is only about 111 (on the conventional 0-10,000 scale), which looks unconcentrated by antitrust conventions. This is not a contradiction: HHI is dominated by the squared shares of only the very largest players, and even Johns Hopkins holds just ~5.6% of the federal total — not enough alone to push HHI high with 674 competitors in the mix. The Gini coefficient captures inequality across the FULL distribution (including the long tail of near-zero recipients), which is the more informative concentration measure here and is why this report leads with it rather than HHI.

How has it changed over time

The warehouse's HERD institution table currently holds only FY2024 (a single year), so the time trend below comes from NCSES's own published trend analysis (fetched directly, ncses.nsf.gov/pubs/nsf26310, "Federal Funding for Science and Engineering: Trends in Rankings Among Top Universities, FYs 2014-23"), which uses NCSES's related Federal S&E Support Survey (a broader survey of federal obligations for science and engineering support, not identical in scope to HERD's R&D-expenditure figure but measuring the same underlying phenomenon):

• FY2014: top 10 recipients received $7.2B of $31.2B total federal S&E obligations to higher education — 22.7%
• FY2023: top 10 recipients received $9.9B of $49.0B total — 20.1%
• FY2023: top 100 recipients received $38.4B of $49.0B — 78.4% (closely matching this report's independently computed FY2024 HERD figure of 81.2% for top 100 — a good cross-survey consistency check)

By the top-10-share measure, concentration among the largest recipients eased slightly over the decade. But NCSES notes two countervailing dynamics: (1) the dollar gap between the #1 recipient (Johns Hopkins, whose total includes its APL research center) and the #2 recipient widened from $764 million in FY2014 to nearly $1.2 billion in FY2023 — the very top of the distribution became MORE unequal even as the top-10 group's overall share shrank; and (2) COVID-19 emergency appropriations (2020-2021) caused temporary but large reshuffling within the top ranks — NYU jumped from 25th place ($450M) in FY2020 to 2nd place ($908M) in FY2021, driven almost entirely by one-time HHS/NIH pandemic-response funding, before rankings mostly reverted to pre-pandemic patterns by FY2022-23. Aside from that pandemic-driven disruption, NCSES describes 'general stability' in which institutions occupy the top 10 over the decade.

Separately, at the national level (not a concentration-across-institutions measure, but relevant context), the FEDERAL SHARE of total academic R&D funding (federal vs. institutional/state/business/nonprofit funding) declined from 59% in FY2013 to a low of 53% (FY2018-2020) before recovering to 55% by FY2023 — a modest reduction in how dependent university research overall is on federal money, distinct from the concentration-across-institutions question this report focuses on.

Data and methodology notes

This warehouse's research.nsf_herd_by_institution table is currently loaded for FY2024 only (verified via data_coverage; the schema declares a 2006-2024 window but only 2024 is actually populated) — a genuine data-ingestion gap that prevented computing the multi-year Gini/HHI trend directly from warehouse data. The time-series comparison (FY2014 vs. FY2023) instead comes from NCSES's own published trend report, fetched and read directly (not a secondary summary). The FY2024 warehouse total ($64.6B federal) is consistent with the separately reported FY2024 total academic R&D figure of $117B at roughly a 55% federal share (~$64B) — a rough cross-check that the warehouse figure is in the right range. The two surveys cited (HERD vs. Federal S&E Support Survey) differ modestly in scope and institution count (674 vs. 1,110), so the FY2024-vs-FY2014/23 comparison in this report is directional, not a single continuous series.

Every query behind this report

3 warehouse calls ran in this session, in order. Each is reproducible against the same snapshot.

SELECT inst_id, institution, SUM(rd_expenditure_usd_thousand) AS federal_total
FROM research.nsf_herd_by_institution
WHERE "year" = 2024 AND funding_source = 'Federal' AND federal_agency = 'Total' AND rd_field = 'All'
GROUP BY inst_id, institution
ORDER BY federal_total DESC
LIMIT 15

SELECT inst_id, SUM(rd_expenditure_usd_thousand) AS federal_total
FROM research.nsf_herd_by_institution
WHERE "year" = 2024 AND funding_source = 'Federal' AND federal_agency = 'Total' AND rd_field = 'All'
GROUP BY inst_id
HAVING SUM(rd_expenditure_usd_thousand) > 0

WITH inst_totals AS (
  SELECT inst_id, SUM(rd_expenditure_usd_thousand) AS federal_total
  FROM research.nsf_herd_by_institution
  WHERE "year" = 2024 AND funding_source = 'Federal' AND federal_agency = 'Total' AND rd_field = 'All'
  GROUP BY inst_id
  HAVING SUM(rd_expenditure_usd_thousand) > 0
),
ranked AS (
  SELECT inst_id, federal_total, ROW_NUMBER() OVER (ORDER BY federal_total DESC) AS rnk
  FROM inst_totals
),
tot AS (SELECT SUM(federal_total) AS grand_total, COUNT(*) AS n FROM inst_totals)
SELECT
  (SELECT SUM(federal_total) FROM ranked WHERE rnk <= 10) AS top10_sum,
  (SELECT SUM(federal_total) FROM ranked WHERE rnk <= 20) AS top20_sum,
  (SELECT SUM(federal_total) FROM ranked WHERE rnk <= 100) AS top100_sum,
  (SELECT grand_total FROM tot) AS grand_total,
  (SELECT n FROM tot) AS n_institutions,
  (SELECT SUM(POWER(federal_total/grand_total*100,2)) FROM ranked, tot) AS hhi