The federal statistical system's decentralized structure is an asset overall, but it creates serious obstacles to coordination and data sharing across agencies and with state partners, according to a new report released by the Bipartisan Policy Center. The report, the second in a four-part series on modernizing federal data infrastructure, draws on 12 months of stakeholder engagement including two roundtables and more than 25 conversations with current and former agency leaders, congressional staff, state officials, private sector professionals, and academic experts. While embedding statistical capacity in mission-focused departments ensures deep subject expertise that a single centralized office would struggle to match, the system now faces two critical problems: fragmented coordination and broken data flows.

Every statistical agency performs the same supporting administrative functions—privacy review, disclosure avoidance, data use agreements, and IT infrastructure—on its own, duplicating effort and missing opportunities for economies of scale. Coordination with states is similarly fragmented: state agencies hold much of the administrative data that federal statistics depend on, such as unemployment insurance wage records, but the methods for producing data and the channels for working with federal partners vary by program. The legal authorities governing data sharing were written in a different technical and administrative era, implementation varies widely across agencies, and privacy frameworks haven't kept pace with re-identification risks made possible by modern computing. The result is a system that often can't share data it's legally permitted to share and increasingly struggles to balance competing demands of access and privacy.

The report proposes establishing shared services across the federal statistical system for common functions, extending data access among the Census Bureau, Bureau of Labor Statistics, and Bureau of Economic Analysis under shared protections, and strengthening data collaboration with and among states. The Bureau of Labor Statistics' Wage Records Program demonstrates what a two-way relationship can look like: BLS compiles quarterly wage records states already collect into a single, longitudinally linked, multi-state dataset and offers participating states information on workers, employers, and jobs in return, with confidentiality protections and no additional reporting burden. Legal restrictions such as Title 13's confidentiality protections limit what Census can share with the other two agencies—for example, the Bureau of Economic Analysis can't directly see the underlying business data that Census collects, even where it would materially improve estimates of gross domestic product and related indicators.

The Advisory Committee on Data for Evidence Building has noted that while comprehensive legislative reform is the ideal end state, administrative streamlining may be more straightforward to achieve in the near term. The report calls for modernizing the legal framework for data sharing and privacy, standardizing data use agreements and disclosure review processes across agencies, and investing in privacy-enhancing technologies and secure access infrastructure. Even where data sharing is legally permitted, agencies currently negotiate data use agreements on an agency-by-agency basis, with significant variation in templates, timelines, and substantive terms—this is why legally permitted sharing often doesn't happen, as the transaction costs are too high. The chief statistician's office, established within the Office of Management and Budget by the 1980 Paperwork Reduction Act, currently operates with only about five full-time positions and has seen more frequent turnover and been filled by noncareer appointments in recent years. Adhering to the office's history of leadership under a nonpolitical, credentialed statistician, alongside modestly bolstered staffing and resources, would allow the chief statistician to better oversee and coordinate these reforms.