Federal Research Funding Has Shifted Away From Life Sciences Toward Physical Sciences, Psychology, and "Other" Fields Over the Past Two Decades

Summary

Over FY2005-FY2024 (the most recent finalized 20-year window in NCSES's field-of-R&D data), total federal obligations for research grew from $53.7B to $98.6B in current dollars — a real (inflation-adjusted) increase of about +14%. But that growth was very unevenly distributed across fields. Life sciences' share of the federal research portfolio fell from 52.3% in FY2005 to 40.8% in FY2024, and in real (CPI-adjusted) dollars life-sciences obligations actually declined about 11% — the only major field to shrink in real terms. Every other field grew faster than the total: physical sciences (+47% real), psychology (+64% real), social sciences (+54% real), math/computer science (+32% real), engineering (+12% real), and a catch-all 'Other' category that more than tripled (+201% real), reflecting the post-2016 rise of newly tracked interdisciplinary/multi-agency research areas (e.g., AI, quantum, biotechnology-adjacent, and pandemic-response supplemental funding not cleanly assignable to one legacy NCSES field). The clearest structural story is a two-decade-long convergence: life sciences' outsized post-1990s dominance (driven by the NIH budget doubling that ran through FY2003) has partially unwound, while physical sciences, math/CS and psychology have closed part of the gap.

Data and method

The connector's own field-of-R&D table (research.nsf_rd_by_field, sourced from the same NCSES Survey of Federal Funds for R&D) only covers FY2016-FY2025 and could not reach back 20 years, so it could not answer the full-window question on its own. Per this corpus's own guidance for a confirmed coverage gap, I fetched NCSES's own underlying data directly via the primary compiler: AAAS's R&D Budget and Policy Program publishes a maintained 'Historical Data.xlsx' workbook (aaas.org/programs/r-d-budget-and-policy/federal-rd-budget-dashboard, last updated 10/23/2025) whose 'Field' sheet is built directly from NCSES Survey of Federal Funds for R&D, Table 9, and runs FY1970-FY2024 in current dollars. That sheet, not a secondary description of it, is the primary data behind every figure in this report. I fetched and parsed that workbook directly (via the connector's web_fetch tool, requesting the 'Field' sheet specifically since the default 5-sheet fetch missed it) and computed FY2005 vs FY2024 changes and shares myself; inflation adjustment to constant 2024 dollars used the connector's own CPI-U-based adjust_inflation tool.

Window choice: 'the last twenty years' as of 2026 spans roughly FY2005-FY2025, but NCSES's field-of-R&D detail is subject to a multi-year publication lag and FY2024 is the most recent year in the AAAS-compiled workbook with a finalized (non-preliminary) field breakdown, so FY2005-FY2024 is the widest fully finalized 20-year window available.

Cross-check: the connector's own research.nsf_rd_by_field table (2016-2025, all-agency, research-obligations-only) shows life sciences at $32.0B in 2016 rising to $38.6B in 2024 (nominal +20.5%), versus the AAAS workbook's $36.3B (2016) to $40.2B (2024) for the same nominal comparison. The two sources broadly agree on the direction (life sciences still growing in nominal dollars over 2016-2024, well below total growth) but differ in level, most likely because the connector's table's post-2021 field taxonomy revision groups a few detailed sub-fields differently than AAAS's own five-decade-consistent bucketing. Both are legitimate cuts of the same underlying NCSES Table 9 data; the AAAS series was used for the headline because only it reaches back to FY2005.

What is NOT covered

This is federal research (basic + applied) obligations only, not experimental development, and not R&D performed with non-federal funds. It captures the field-of-science allocation across ALL federal funding agencies and ALL performers (universities, federal labs, industry, nonprofits) combined — a broader population than the connector's own university-only HERD table would have given. It does not break out by performing sector or by individual agency (e.g., how much of the life-sciences shift is NIH vs. other agencies) — that would require the AAAS workbook's separate 'Agency' and 'Discipline' sheets, which were not analyzed for this report.

Every query behind this report

1 warehouse call ran in this session, in order. Each is reproducible against the same snapshot.

SELECT year, rd_field, obligations_usd_million FROM research.nsf_rd_by_field WHERE field_level = 1 AND year IN (2016, 2024) ORDER BY year, rd_field