Scientific Research and Development Services

NAICS 2022 code 5417 — United States (industry-group rollup primer)

---

1. Overview

This is the United States' research-and-development-for-hire industry group — the establishments whose main line of business is doing research and development (R&D) as a paid service, rather than making a product. When a drug company, a defense agency, a foundation, a state government, or a venture-backed startup wants knowledge invented, tested, and de-risked but does not want to build and staff the lab itself, the work flows to firms in this group. The North American Industry Classification System (NAICS) code 5417 is the "industry group" (4-digit) level that gathers two very different research businesses under one heading: the hard-science shop (54171) — physical, engineering, and life-sciences R&D — and the social-science-and-humanities shop (54172).[1]

The single most important thing to understand at this level is what it leaves out. R&D is usually a cost center inside another company or institution. When an automaker, a chip maker, or a big drug firm runs its own labs, that spending is booked under its industry — auto manufacturing, semiconductor manufacturing, pharmaceutical manufacturing — not here. When a university or a national lab performs research, it is counted under education or public administration. This group captures only the slice of American research that is bought and sold as a stand-alone service: the merchant R&D market. That makes it a much smaller number than "how much R&D America does," but a cleaner window into a distinct, high-wage services economy.[6]

The distinctive value of looking at the group rather than the two codes underneath it is the contrast — and here the contrast is unusually stark. This group is a whale and a minnow bolted together: one child is roughly 97% of the money and the other about 3%, and the two differ in almost everything that matters to an investor — what they study, who pays them, who owns them, how much they pay their people, and whether you can buy them at all. That asymmetry is Section 2.

---

2. What's inside — the two children and how they differ

The group has exactly two children. 54171 — Research and Development in the Physical, Engineering, and Life Sciences is itself a large industry containing nanotechnology, biotechnology, and a big "everything-else" catch-all of contract-research and engineering/defense/energy labs.[4] 54172 — Research and Development in the Social Sciences and Humanities is a single-industry group covering economics, sociology, psychology, education, public-health behavior, and policy research.[5] They share one economic DNA — selling credentialed technical labor and keeping it billable on funded projects — but they are near-opposites in scale, buyer base, and investability.

Contrast table

Acronyms: HHI = Herfindahl-Hirschman Index, a 0–10,000 market-concentration gauge (regulators treat ~1,500+ as "concentrated"); CR4 = combined revenue share of the four largest firms; CRO = contract research organization; VC = venture capital; PE = private equity; ETF = exchange-traded fund.[3][4][5]

How to read the contrast. Three patterns dominate:

1. One child is the group. 54171 is roughly nineteen dollars of every twenty in receipts and nineteen jobs of every twenty. Whatever is true of the physical/engineering/life-sciences business is approximately true of NAICS 5417 as a whole; the social-science child barely moves the aggregate. Read the group's headline numbers as a hard-science story with a small policy-research footnote.[3]

2. The buyers are different, so the cycles are different. 54171's swing factor is the capital and corporate R&D cycle — biopharma budgets, venture funding, the drug patent cliff, federal science appropriations. 54172's swing factor is almost purely the federal budget and the politics around it. In 2025 those diverged sharply: the hard-science side kept its structural tailwinds while the social-science side absorbed contract terminations and grant cancellations (Sections 6, 9).[4][5]

3. Investability is lopsided. 54171 has real, if uneven, public doors — biotech funds, the listed CRO oligopoly, federal-services contractors. 54172 has essentially none: no pure-play listed company exists, its center of gravity is non-profit institutes that cannot be bought at any price, and the only listed proxies embed the work inside much larger consultancies. Nearly all genuine ownership at this level lives in 54171.

---

3. How big it is (this group's rollup figures)

Our ground-truth federal figures for the whole 5417 group. They come from two Census programs and two different years, so treat them as a set of readings, not one continuous income statement.[2][3]

The rollup reconciles cleanly. The two children's employment adds up exactly to the group total (957,341 + 46,634 = 1,003,975), their 2022 receipts sum to the group's $340.5 billion ($331.2B + $9.30B), their establishments sum to 22,403, and their payroll sums to ~$154 billion — a strong integrity check that the children are mutually exclusive slices of the same whole. The one small exception is firm counts: the children sum to ~18,232 firms against the group's 18,219, because a single firm operating establishments in both children is counted once at the parent level.[3]

The pooled group is even more fragmented than either child. Its HHI of 59 sits below both 54171's 62 and 54172's 390, and its top-four firms hold just 11.1% of receipts — because combining two different research fields dilutes any one firm's share. There is no dominant company anywhere in this group.[3]

This is a high-wage industry group — but the average hides a split. ~$154 billion of payroll across ~1.0 million workers is roughly $153,500 per employee, a workforce of PhD scientists, engineers, clinicians, statisticians, and analysts. But that blended figure is really 54171's ~$156,000 pulled down slightly by 54172's ~$92,000; the hard sciences pay materially more than the social sciences.[2]

The undercount caveat is large — and it is worse for the social-science child. The $340.5 billion of receipts counts only R&D sold as a service. It is a small fraction of all research performed in the United States, for several reasons:

• In-house corporate R&D is elsewhere. A carmaker's, chip maker's, or drug company's own labs are classified under the parent's manufacturing industry, not here — the biggest gap by far.[4]
• Universities and government/national labs are excluded — university research sits under education, federal labs under public administration — even though together they perform an enormous share of U.S. science. This exclusion is especially severe for 54172, because most federally funded behavioral and social research is actually performed inside universities.[5]
• County Business Patterns omits the self-employed, no-payroll operators, and most government workers, so the fat tail of one- and two-person specialist shops (dense in both children) is under-counted.[2]
• Equity-funded, pre-revenue firms report little or no "receipts" even while employing hundreds — a clinical-stage biotech living on investor capital barely registers in a revenue-based measure.[3]

For scale: total U.S. R&D reached about $937 billion in 2023, of which businesses performed roughly $722 billion — the vast majority in-house and booked under other industry codes.[6] Read the $340.5 billion here as "the market for R&D you can hire someone else to do," not as the nation's research budget.

---

4. The investable universe — where value concentrates across the children

There is no pure public play on the whole group, and essentially all of the listed exposure that does exist sits in the hard-science child. Where you can put money differs enormously between the two. (Tickers and scale figures appear here and in Section 10 only.)

54171 — the deep, uneven public and private pool (≈97% of the group). This is where nearly all the investable activity lives, and it splits into recognizable clusters:

• Biotechnology is the most public: hundreds of listed clinical-stage names (e.g., Alnylam/ALNY, CRISPR Therapeutics/CRSP, Recursion/RXRX), reached most simply through two ETFs — XBI (equal-weighted, closest to the research layer) and IBB (cap-weighted). The largest "biotech" names everyone knows (Eli Lilly/LLY, Amgen/AMGN, Vertex/VRTX) are technically pharmaceutical manufacturers outside the code.[4]
• The contract-research / engineering / federal-services catch-all offers two clean listed clusters: the clinical-CRO complex (IQVIA/IQV, ICON/ICLR, Medpace/MEDP, Charles River/CRL, Fortrea/FTRE) for R&D outsourcing, and federal science-and-engineering contractors (Leidos/LDOS, SAIC, Amentum/AMTM, Parsons/PSN, KBR, Jacobs/J, Exponent/EXPO) for the physical-sciences half.[4]
• Nanotechnology is nearly impossible to own directly — the credible route is indirect "picks-and-shovels" (Applied Materials/AMAT, Lam Research/LRCX, KLA/KLAC, Bruker/BRKR).[4]
• The strategic anchors — non-profit institutes such as Battelle, SRI International, and MITRE — cannot be bought at any price, and much of the mid-tier CRO layer is PE-owned (Parexel, Syneos).[4]

54172 — almost nothing to own (≈3% of the group). No listed company is dedicated solely to social-science and humanities research. The only practical listed exposure is indirect and thematic: ICF International (NASDAQ: ICFI, ~$2.0B revenue) for social-program evaluation and survey work, and Charles River Associates (NASDAQ: CRAI, ~$687M revenue) for antitrust, damages, and regulatory economics — in both, this research is one line among many.[5] The real center of gravity is non-profit or closely held and not investable as equity: RAND, NORC, Westat, Mathematica, Abt Global, the American Institutes for Research (AIR), RTI International, and think tanks such as the Urban Institute and Brookings. Nine of these policy-research leaders alone employ more than 20,000 people — roughly half the entire measured group's social-science workforce, underscoring the undercount in Section 3.[5]

Bottom line for the reader. If you want listed exposure to NAICS 5417, you are effectively buying into 54171 — biotech ETFs and the CRO / federal-services complex are the realistic doors. The social-science child is reachable in public markets only through two diluted proxies. Across the whole group, the deepest and most genuine exposure is private and long-dated.

---

5. How the money works

At every corner of this group the business is people-and-projects: the cost base is overwhelmingly salaries for credentialed researchers, and margin comes from keeping that talent on paid, funded work. Idle PhDs are how these firms lose money. But the two children run on different financial machinery:

• 54171 layers capital and long-cycle payoffs on top of labor. Lab-heavy segments carry expensive equipment. Service-heavy CROs and engineering shops sell multi-year contracts and watch bookings, backlog (signed-but-undelivered work — large but cancellable), book-to-bill ratio (above 1.0 means the order book is growing), and billable utilization. Early-stage nano and biotech firms live on milestones, licensing, and royalties — the high-margin, slow-arriving payoff — and on non-dilutive grants. The defining reality is the "valley of death," the long cash-hungry gap between a promising result and a paying product; for pre-revenue firms the number that matters most is cash runway versus burn rate.[4]
• 54172 is lighter on capital and runs on cost recovery. The core mechanic is billing direct labor and recovering overhead: a researcher's hours bill to a project, and the firm layers on negotiated indirect cost rates (for non-profits, a "facilities-and-administrative," or F&A, rate) to recover shared expenses. The levers are utilization, indirect-rate competitiveness, backlog/book-to-bill, and win/renewal rates. Margins split by mix — the pure-government, cost-reimbursement end is low-margin and steady (ICF runs ~11% adjusted-EBITDA — earnings before interest, taxes, depreciation, and amortization — margin), while commercial/litigation economics is higher-margin and cyclical with the economy rather than the federal budget.[5]

The common thread: this is an expensive fixed cost base of skilled labor, and financial health at both children turns on keeping that talent funded and billable.

---

6. What drives demand

Demand for this group is really demand for outsourced research and for the capital and appropriations that fund it — and the two children answer to different masters:

• 54171 tracks the corporate and federal science cycle. Big pharma keeps shifting development to CROs (large-pharma outsourcing is estimated near ~45% and rising); large drugmakers face roughly $170 billion of revenue exposed to patent expirations by 2032, creating urgent demand to acquire biotech pipelines; biotech-trial funding is interest-rate-sensitive (the 2021–23 slump and 2025 recovery tracked the cost of capital); and federal science budgets are large — in fiscal 2024, five agencies (Defense, Health and Human Services, Energy, NASA, and the National Science Foundation) accounted for ~93% of federal R&D obligations, roughly $181 billion. Scientific waves — advanced semiconductors and AI hardware, obesity/metabolic drugs, cell and gene therapy — pull in both spend and capital.[4]
• 54172 tracks the federal policy budget, almost alone. The dominant buyers are the National Science Foundation's Social, Behavioral and Economic Sciences directorate, the Institute of Education Sciences (IES), the National Institutes of Health's behavioral research, and the National Endowment for the Humanities (NEH), reinforced by evidence-based-policy mandates. Because government dominates, aggregate demand is politically cyclical; the resilient counter-cyclical slice is commercial/litigation economics (antitrust and damages work).[5]

Looking forward, artificial intelligence (AI) cuts across both — automating routine literature review, coding, and reporting while raising the premium on what it cannot fake: credible samples, causal inference, defensible methods, and lab-validated results.[4][5]

---

7. Regulation

There is no single "R&D law." Oversight attaches to the work being done, and the two children face different regulatory stacks:

• 54171 — clinical, chemical, and procurement rules. Good Clinical Practice and Good Laboratory Practice under the Food and Drug Administration (FDA) govern human trials and nonclinical safety studies; the FDA's approval gate is the main value-creating event for a drug program. Nanomaterials are regulated product-by-product (Environmental Protection Agency reporting under the Toxic Substances Control Act, FDA pathways, worker-exposure guidance). Federal work adds the Federal Acquisition Regulation (FAR) Part 35 on R&D contracting, export controls, and the Bayh-Dole framework on inventions. A live geopolitical rule is the BIOSECURE Act (signed December 2025), which restricts federal dealings with named Chinese biotech "companies of concern" — a reshoring tailwind for U.S. providers.[4]
• 54172 — contracting, ethics, and data rules. FAR Part 31 and the Office of Management and Budget's Uniform Guidance (2 CFR 200) govern allowable and indirect costs; the Common Rule (45 CFR 46) requires Institutional Review Board (IRB) review and informed consent for human-subjects research; and data-privacy law — FERPA (education records), HIPAA (health data), and CIPSEA (federal statistical confidentiality) — plus Paperwork Reduction Act clearance for federal surveys. The most acute live issue is the fight to cap federal indirect-cost recovery (Section 9).[5]

Both children share exposure to human-subjects ethics and federal-contracting rules; the divergence is that 54171 answers additionally to the FDA and export regime, while 54172's binding pressure point is overhead-cost policy.

---

8. Consolidation

The group is fragmented in aggregate and consolidating in pockets — a classic barbell.

• The research layer is extraordinarily un-concentrated. At the group level the top four firms hold just 11.1% of receipts and the HHI is 59 — far below the ~1,500 "concentrated" threshold — reflecting more than 18,000 firms across two fields.[3]
• But sub-markets consolidate hard, mostly inside 54171. The clinical-CRO tier is an oligopoly built by big deals — Thermo Fisher bought PPD for $17.4 billion (2021), ICON bought PRA for ~$12 billion (2021), Syneos went private for ~$7.1 billion (2023). Biotech's endgame is acquisition, not organic scale: driven by the patent cliff, 2025 saw well over $200 billion of pharma M&A (mergers and acquisitions), with large drugmakers buying de-risked pipelines rather than building them.[4]
• 54172 consolidates slowly. PE roll-ups of government-services firms and university spin-outs are real but gradual, and the tax-exempt segment structurally limits classic M&A because non-profit institutes cannot simply be bought.[5]

Across the whole group, the realized return driver is usually an acquisition or a licensing royalty, not an initial public offering (IPO), and the non-profit institutes in both children are a permanent, unbuyable fixture that shapes competition without ever changing hands.[4][5]

---

9. Risks

• Funding-cycle dependence — and it diverges by child. 54171's grant-reliant and clinical-stage firms are exposed to federal science budgets and rate-sensitive biotech financing; 54172 is exposed almost entirely to the federal policy budget, and that risk materialized in 2025 — government-efficiency contract reviews terminated large volumes of research work (Mathematica and AIR each lost 15 IES agreements), the NEH cancelled more than 1,200 grants, and the NSF social-science directorate faced a fiscal-2026 request cutting it to roughly a third of its prior level.[4][5]
• Indirect-rate cap risk (54172 especially). A 2025 attempt to cap NIH indirect (F&A) reimbursement at 15% — versus negotiated rates averaging ~58% — was blocked in court but signals durable pressure to compress overhead recovery, a direct hit to the economics in Section 5.[5]
• The valley of death (54171). Many projects work in the lab and never survive scale-up, clinical failure, or qualification; in biotech only ~1 in 10 drugs entering first-in-human trials wins approval.[4]
• Cancellable backlog and buyer concentration. A single sponsor or agency can scrap a program overnight; a consolidating pharma client base concentrates leverage on CROs, and 54172's dependence on one dominant buyer (Washington) amplifies every budget swing.[4][5]
• Capital intensity and dilution (54171). Pre-revenue micro-caps and private firms burn cash and raise equity repeatedly.[4]
• Talent-cost inflation and AI commoditization. Wage inflation among cleared scientists and PhDs compresses margins; AI threatens to commoditize routine analytic work (a sharper risk for 54172's reporting-heavy tasks).[4][5]
• Reputational, IP, and quality failure. Credibility is the product — an FDA warning letter, a data-integrity finding, or a contested/non-reproducible study can cost re-competes and program value.[4][5]
• Measurement and "washing" risk. Federal business statistics omit government, university, and self-employed research, and some firms brand work as "nano" or "AI" without material technology. Do not mistake broad sector exposure for direct ownership of this group's activity.[2]

---

10. How to invest and the outlook

Public routes — and they are almost entirely 54171.

• Physical/engineering/life sciences (54171): biotech ETFs are the standard entry — XBI (equal-weight, closest to the research layer) and IBB (cap-weighted); the CRO complex (IQV, ICLR, MEDP, FTRE, CRL) for the cleanest R&D-outsourcing exposure; federal science-and-engineering contractors (LDOS, SAIC, AMTM, PSN, KBR, J, EXPO); and, for nanotechnology, indirect "picks-and-shovels" instrument and equipment makers (AMAT, LRCX, KLAC, BRKR). These trade on backlog growth, book-to-bill, and margins rather than dividends.[4]
• Social sciences and humanities (54172): no pure play. Practical listed exposure is limited to ICF International (ICFI) for government program evaluation and Charles River Associates (CRAI) for litigation and regulatory economics — both thematic, not pure, because most of their revenue sits outside this code; CRAI's commercial mix makes it the more insulated of the two from the 2025 federal-budget shock.[5]

Private routes are the deepest and most genuine exposure in both children — VC and life-science funds, growth equity, corporate-venture partnerships, university technology transfer, royalty and venture-debt strategies, and PE ownership of the mid-tier CROs and government-services platforms. The trade-off is illiquidity and long horizons; the realistic payoff is an acquisition or a licensing royalty rather than a near-term IPO. Core diligence everywhere: intellectual-property ownership and freedom to operate, customer/agency validation, cash runway versus burn, contract quality and backlog funding, and indirect-rate competitiveness.[4][5]

Forward-looking judgment (not a reported fact). Because 54171 is ~97% of the group, NAICS 5417's outlook is essentially 54171's — a quality-cyclical, high-skill services industry with real structural tailwinds (rising outsourcing penetration, BIOSECURE-driven reshoring, AI adoption in trial and research design, steady if budget-exposed federal science demand), inside which the biotech corner remains one of the highest-beta parts of the market. The social-science child (54172) is small enough not to move the aggregate, but it is the group's clearest cautionary signal: its near-total dependence on the federal policy budget turned into an outright shock in 2025, and it will stay under pressure until the appropriations fight resolves — even as the underlying demand for evidence remains intact. Net: treat 5417 as a large, fragmented, high-wage merchant-research industry whose growth is real but geared to the biopharma, capital, and government funding cycles — and never mistake its $340.5 billion of receipts for the total amount of R&D the United States performs, or assume any listed company's full revenue belongs to this group.

---

Sources

1. U.S. Census Bureau, "2022 NAICS — 5417 / 54171 / 54172: Scientific Research and Development Services" (industry-group definition, scope, and children). https://www.census.gov/naics/?input=5417&year=2022
2. U.S. Census Bureau, County Business Patterns, NAICS 5417 and children (establishments, employment, annual and Q1 payroll; coverage exclusions of government, non-employer, and self-employed), 2023. Ground-truth group figures per Histometrics ingested stats (22,403 establishments; 1,003,975 employees; $154.07B annual payroll; $44.80B Q1 payroll). https://www.census.gov/programs-surveys/cbp.html
3. U.S. Census Bureau, 2022 Economic Census — Concentration by Largest Firms, NAICS 5417 and children (receipts, firm counts, CR4/CR8/CR20/CR50, HHI). Ground-truth group figures per Histometrics ingested stats (receipts $340.52B; 18,219 firms; CR4 11.1% / CR8 16.8% / CR20 28.3% / CR50 42.3%; HHI 59). https://www.census.gov/programs-surveys/economic-census.html
4. Histometrics primer, NAICS 54171 — Research and Development in the Physical, Engineering, and Life Sciences: nanotechnology / biotechnology / catch-all children, biotech ETFs (XBI/IBB) and clinical-stage universe, CRO and federal-services complex, ~$38B biotech VC (2025), ~$170B+ patent cliff by 2032, 2025 pharma M&A (>$200B), ~45% large-pharma outsourcing, federal R&D obligations (~$181B / 93% across five agencies, FY2024), NNI budget, FDA/FAR/TSCA regulation, and the BIOSECURE Act (Dec. 2025). Draws on U.S. Census, NSF NCSES, FDA, FAR, and SEC filings.
5. Histometrics primer, NAICS 54172 / 541720 — Research and Development in the Social Sciences and Humanities: pass-through single-child group; ICF International (ICFI, ~$2.0B revenue, ~11% adj-EBITDA margin) and Charles River Associates (CRAI, ~$687M revenue) as thematic listed proxies; non-profit institutes (RAND, NORC, Westat, Mathematica, Abt, AIR) and >20,000 employees across nine policy-research leaders; FAR Part 31 / OMB Uniform Guidance / Common Rule / FERPA-HIPAA-CIPSEA regulation; 2025 federal-budget shock (IES terminations, NEH grant cancellations, NSF SBE fiscal-2026 request cut to ~1/3, NIH 15% indirect-cap attempt blocked in court). Draws on U.S. Census, NSF, IES, NEH, NIH, and SEC filings.
6. National Center for Science and Engineering Statistics (NSF NCSES), "U.S. R&D Totaled $937 Billion in 2023; Business R&D $722 Billion," 2025. https://ncses.nsf.gov/pubs/nsf26314