In September 2017, the bipartisan Commission for Evidence-based Policymaking released twenty-two recommendations to improve secure data access for evidence building activities involving population-level government files. Many of the files are siloed in government agencies. The commission deliberated over eighteen months to understand the risks and barriers to broader data use.
Objectives and Approach
I will describe the Commission’s charge and review its recommendations, in context of US laws and privacy debates. I will compare the report’s recommendations and implications to laws and initiatives in other countries. The report calls for the establishment of a National Secure Data Service (NSDS), which has the potential to transform the data sharing environment for federal agencies, policy makers, and researchers. The report suggests more extensive use of differential privacy and secure multiparty computation to protect privacy. I will describe how the current environment could change depending on how the recommendations are implemented.
The Commission was established under one administration, but the recommendations were released under another. Despite the political and budget uncertainty in Washington, a bill was introduced and passed in the House in November 2017 to implement some recommendations. I will summarize the actions to be taken if the bill becomes law, including directives on learning agendas to prioritize and coordinate evidence-building activities across government, the roles of chief evaluation and chief data officers, and formation of an advisory committee to plan a NSDS. I will describe benefits that could follow from directives in the bill, including transparency about uses of administrative data, development of guidance to assess the risk when combining data sources, and minimization of the risk of publicly releasing de-identified data.
The US may develop a national secure data service to support evaluations and policymaking. The recommendations are akin to the UK Data Service. Some recommendations are straightforward, others need years of planning and technical breakthroughs, and all require political buy-in and funding.