The Federal Statistical System — 13 Principal Agencies & Independence Norms

The United States federal statistical system is deliberately decentralized — 13 principal statistical agencies spread across 11 departments, each embedded within a mission agency whose policy work the statistics are designed to inform. This is almost unique in the developed world: most peer countries (the UK, Canada, France, Germany, Australia) centralize all national statistics in a single independent office. The U.S. approach emerged from the Progressive Era belief that statistics embedded within mission agencies would be more practically useful and more adequately funded. The tradeoff is chronic coordination challenges, inconsistent methodologies across agencies, and statistical agencies that are only partially shielded from the political priorities of their parent departments. In 2025, the DOGE initiative's attempts to access confidential statistical microdata — raising questions about Statistical Policy Directive No. 1 and the Confidential Information Protection and Statistical Efficiency Act (CIPSEA) — put these long-standing independence norms under their most direct stress in decades.

Legal Authority

• 44 U.S.C. §§ 3501 et seq. — Paperwork Reduction Act: establishes OMB's role in coordinating federal statistical activities through the Office of the Statistical and Science Policy; authorizes OMB to set standards for statistical activities across agencies
• 44 U.S.C. §§ 3561–3583 — CIPSEA (Confidential Information Protection and Statistical Efficiency Act): prohibits federal agencies from using statistical data (collected under a promise of confidentiality for statistical purposes) for non-statistical purposes — including law enforcement, regulatory, or administrative actions; the legal protection at issue when DOGE sought access to IRS and SSA microdata in 2025
• Statistical Policy Directive No. 1 — OMB directive establishing professional independence norms for federal statistical agencies: prohibits political interference in the preparation and release of statistical data; the norm under which political officials are barred from receiving advance access to CPI, employment, or GDP releases before public release

Key Mechanics

The U.S. federal statistical system is decentralized across 13 principal statistical agencies — BLS (labor), BEA (economic accounts), Census (population and economic census), BTS (transportation), NCHS (health), NASS (agriculture), and others — each embedded in a mission department. This decentralization means coordination depends on OMB's statistical policy function rather than a single central authority. CIPSEA's confidentiality protection is the primary legal shield for statistical data: when an agency collects data under a pledge of statistical confidentiality, that data cannot be used for non-statistical purposes regardless of executive directive. Statistical Policy Directive No. 1 enforces the release norm: major economic indicators (CPI, jobs, GDP) are released at the same time to all users, with no advance briefings for political officials — a protocol that maintains market integrity and public trust in the data.

Structure of the System

The 13 Principal Statistical Agencies

The Office of Management and Budget (OMB) formally designates "principal statistical agencies" — agencies whose primary mission is the collection, processing, and dissemination of statistics for general public use rather than for a specific regulatory or operational purpose. The 13 are:

1. Bureau of Labor Statistics (BLS) — Dept. of Labor; employment, wages, prices (CPI), productivity
2. Bureau of Economic Analysis (BEA) — Dept. of Commerce; GDP, national accounts, PCE price index
3. Census Bureau — Dept. of Commerce; decennial census, ACS, economic censuses, population estimates
4. Bureau of Justice Statistics (BJS) — Dept. of Justice; crime victimization, courts, corrections
5. Bureau of Transportation Statistics (BTS) — Dept. of Transportation; freight, aviation, safety
6. National Agricultural Statistics Service (NASS) — Dept. of Agriculture; crop production, livestock, farm economics
7. Economic Research Service (ERS) — Dept. of Agriculture; food security, farm income, rural economics
8. National Center for Health Statistics (NCHS) — HHS/CDC; vital statistics, NHANES, NHIS
9. National Center for Education Statistics (NCES) — Dept. of Education/IES; NAEP, IPEDS, CCD
10. Energy Information Administration (EIA) — Dept. of Energy; energy production, consumption, prices, reserves
11. National Center for Science and Engineering Statistics (NCSES) — NSF; R&D spending, STEM workforce
12. Statistics of Income Division (SOI) — IRS/Treasury; tax return data, corporate income, estate taxes
13. Office of Research, Evaluation, and Statistics (ORES) — SSA; Social Security program statistics

Beyond these 13, dozens of other federal agencies produce significant statistical output without principal agency designation — including the Federal Reserve (financial accounts), HUD (housing finance), USCIS (immigration), and the Defense Manpower Data Center (military personnel).

Independence Norms: Statistical Policy Directive No. 1

The cornerstone of U.S. statistical agency independence is OMB Statistical Policy Directive No. 1 (SPD 1), first issued in 1985 and revised in 2014. SPD 1 establishes that principal statistical agencies must:

• Protect the integrity of the statistical process — political appointees cannot alter methodologies, release timing, or data access based on political considerations
• Prohibit prepublication review by non-statistical officials — agency heads and political appointees cannot see statistical releases before public release time; embargoed data is shared only with specifically authorized statistical officials
• Maintain public trust in statistical objectivity — methodologies must be documented and transparent; revisions must follow stated procedures

SPD 1 is administrative policy, not law. An administration could theoretically override it. The independence norm has held across administrations primarily because the financial markets' dependence on the data creates powerful private-sector constituencies for its integrity — Wall Street's ability to trade on BLS and BEA releases depends on their credibility.

The most notable pre-SPD 1 violation of statistical independence occurred in 1971, when the Nixon administration pressured BLS Commissioner Harold Goldstein over unemployment data; Goldstein resigned rather than alter the data. SPD 1 was subsequently developed to formalize protections against such interference.

CIPSEA: Confidential Data Protections

The Confidential Information Protection and Statistical Efficiency Act of 2002 (CIPSEA), codified at 44 U.S.C. § 3561–3583, provides the legal basis for statistical agency confidentiality protections and data-sharing authorities:

Confidentiality protections (Part I): Statistical data collected under a pledge of confidentiality — meaning respondents were told their information would be used only for statistical purposes — cannot be disclosed in identifiable form to anyone outside the collecting agency, including law enforcement, regulatory agencies, or other statistical agencies, without the respondent's consent. Violations carry criminal penalties. This applies to survey microdata, administrative data collected for statistical purposes, and data obtained from other agencies under statistical data-sharing agreements.

Data-sharing authorization (Part II): CIPSEA authorizes statistical agencies to share business data with one another for statistical (not regulatory) purposes — the legal basis for Census-BLS-BEA data integration that underlies the Longitudinal Business Database and related administrative data products.

CIPSEA II (2018): The Foundations for Evidence-Based Policymaking Act strengthened CIPSEA by expanding authorized data-sharing for research and evaluation purposes while maintaining the prohibition on disclosure for non-statistical uses. It also established the National Secure Data Service pilot (see federal-data-infrastructure).

DOGE and CIPSEA Confrontation (2025)

The 2025 DOGE initiative created the most direct confrontation with CIPSEA protections since the statute's enactment. DOGE sought access to:

• BLS microdata linking individual survey responses across CPS, CES, and QCEW files
• IRS SOI Division taxpayer return data (additionally protected by 26 U.S.C. § 6103)
• SSA records including wage histories and beneficiary information
• HHS/CMS claims data from Medicare and Medicaid

Statistical agency leaders argued that DOGE access would violate CIPSEA's prohibition on non-statistical use of pledged-confidentiality data, and that unauthorized disclosure could undermine survey response rates by destroying respondent trust — a self-defeating outcome for the statistical system as a whole. Federal courts issued temporary restraining orders on some data access requests in early 2025 based on Privacy Act and CIPSEA grounds.

The Evidence-Based Policymaking Framework

The Foundations for Evidence-Based Policymaking Act of 2018 (Evidence Act) created a new data governance architecture across all CFO Act agencies:

• Chief Data Officers (CDOs): each CFO Act agency must designate a CDO responsible for data governance, quality standards, data inventory, and data-sharing facilitation
• Statistical Officials: each agency designates a statistical official as liaison to OMB statistical standards
• Evaluation Officers: independent program evaluation capacity at each agency
• Learning Agendas: four-year evidence-building plans tied to agency strategic plans (5 U.S.C. § 306), identifying priority policy questions and evaluation methods
• Federal Data Service (FDS): the long-term vision for a secure, privacy-preserving research environment enabling administrative data linkages — piloted at Census as the National Secure Data Service

International Comparison

The U.S. decentralized model is the global outlier among peer economies:

The decentralized U.S. model produces some advantages: statistical agencies embedded in mission departments have deeper subject-matter expertise and closer ties to the data-generating administrative processes. NASS's deep integration with USDA agricultural programs, for example, gives it access to farm data that a standalone statistics office might not obtain. The disadvantages: higher total cost (each agency maintains its own IT infrastructure, collection systems, and review processes), inconsistent methodological standards, and the recurring coordination challenge of building integrated statistics across agency lines.

How It Affects You

If you are a citizen or consumer: The statistical system's outputs determine billions in government transfers you may rely on — Social Security COLAs are set by BLS CPI-W; Medicaid FMAP rates use BEA state income data; Special Education grants use Census population data. The quality and independence of these agencies is not abstract: politically manipulated statistics have historically preceded economic crises in countries where independence protections are weak.

If you are a business, researcher, or analyst: Statistical agencies publish much of their data free — BLS, BEA, Census, and EIA APIs are publicly available with no cost. Researchers can access restricted microdata through Federal Statistical Research Data Centers (FSRDCs) — a network of 30+ secure computing sites at universities where approved researchers can access confidential survey and administrative data under CIPSEA protections without the data leaving secure federal systems.

If you work at a federal agency: The Evidence Act's requirements — Chief Data Officers, Learning Agendas, Evaluation Officers — are statutory mandates, not recommendations. OMB SPD 1 sets the floor for how statistical releases must be handled. If your agency collects data under a confidentiality pledge, CIPSEA prohibits sharing that data with law enforcement, regulatory offices, or DOGE-equivalent bodies without respondent consent.

If you are a journalist or policy analyst: The 13-agency structure creates coverage gaps — no single source covers the whole system. OMB's annual "Statistical Programs of the United States Government" report (published on OMB.gov) is the most comprehensive overview of all federal statistical activities and budgets. When covering DOGE-related data access disputes, the key legal hook is CIPSEA § 3561–3583 and the distinction between "statistical" and "non-statistical" uses of pledged-confidentiality data.

Recent Developments

• 2025 — DOGE sought access to multiple statistical agency microdata files; courts issued TROs based on CIPSEA and Privacy Act grounds; OMB SPD 1 was invoked by agency career leaders; no statistical releases were altered or delayed.
• 2024 — Evidence Act Learning Agendas completed first 4-year cycle across CFO Act agencies; OMB published review of implementation gaps.
• 2022 — FedRAMP Authorization Act codified cloud security standards applicable to statistical agency data infrastructure.
• 2018 — Foundations for Evidence-Based Policymaking Act enacted; created CDO Council, CIPSEA II, and National Secure Data Service authorization.
• 2002 — CIPSEA enacted (P.L. 107-347, Title V); first comprehensive federal statistical confidentiality statute.
• 1985 — OMB Statistical Policy Directive No. 1 issued; formalized independence norm prohibiting political interference with statistical releases.