Learning More from the Data the Federal Government Already Collects

Last week the Departments of Labor and Education published the draft regulations implementing the Workforce Innovation and Opportunity Act (WIOA).  These proposed regulations, and the underlying WIOA legislation, will improve our Nation’s public workforce system by strengthening coordination and accountability. One of their most exciting features is that they would require States to produce standardized, easily-understandable “scorecards.” What that means is that – for the first time – workers choosing among different training programs that receive WIOA funding will be able to easily compare them on criteria that matter, like how much the program costs, the percent of participants who actually complete the program, and the average earnings of participants. Right now, workers seeking training may not even be able to find apples-to-apples data on how much different programs would cost, much less how their students fare in the job market. WIOA and the proposed regulations published last week seek to change that. In addition, the draft regulations put in place the WIOA requirement that States implement high-quality evaluations of the core WIOA employment, education, and training programs. That will let them improve workforce outcomes over time by learning which approaches work best and then scaling-up those approaches.

All of this is possible because the proposed regulations encourage States to leverage their “administrative data” for maximum effect. Administrative data are data that are already collected by government entities for program administration, regulatory, or law enforcement purposes. In this context, administrative data include, for example, participant enrollment information and State-held unemployment insurance (UI) records on employment and earnings.  Being able to match these data is essential for States to reliably and cost-effectively generate accurate information about the employment outcomes that employment, education, and training programs achieve. It’s equally essential that States conduct these matches while protecting the privacy or confidentiality of these data, and that we ensure that personally identifiable, individual-level data are not disclosed to private entities without an individual’s consent.  The proposed regulations provide clarity on the structures that States can use to do that: match participant enrollment information, including information from training providers, with State-held UI employment and earnings records while maintaining the privacy and confidentiality.  

Beyond job training, making better use of the administrative data the Federal Government already collects has huge potential to provide the public with better information and help government agencies learn which approaches work best so that they can further improve government programs.  Often, using administrative data lets us produce more reliable information at a lower cost than if government relied solely on more expensive and time-consuming data collection methods, like surveys.  Combining administrative and survey data can be especially fruitful. For example, multiple studies on student aid simplification showed the feasibility and importance of simplifying the Free Application for Federal Student Aid (FAFSA), using Federal administrative records and survey data together.  This research influenced the steps the Administration has already taken to simplify the FAFSA and motivated both Administration and Congressional proposals to make further legislative progress.

Administrative data can undoubtedly be an incredibly powerful tool to help government identify and do more of what works. Unfortunately, there are some significant barriers to making better use of these data, including out-of-date statutes that have not been updated for modern technology and data analytic techniques, and resource and capacity constraints that prevent agencies from making their data usable for research or other statistical analysis.

The President’s 2016 Budget includes a package of proposals that would begin to address these challenges and help make additional administrative data from Federal agencies and programs legally and practically available for policy development, program evaluation, performance measurement, and accountability and transparency purposes. All of these proposals would operate within a strong framework of privacy, confidentiality, and data security protections.  The President is calling for:

• Expanding access to data. Consistent with bipartisan congressional proposals, the Budget would improve access to employment and earnings data by letting select Federal statistical and evaluation units access the Department of Health and Human Services’ (HHS) National Directory of New Hires (NDNH) database—a national database of new hire, quarterly wage, UI claim information—for statistical purposes. A limited number of entities have authority to receive matched NDNH data for limited purposes, but stringent access restrictions prevent the use of these data for evaluations that aim to measure the impact of federally-funded job training programs, even though the Department of Labor administers the UI system. This proposal would allow NDNH data to be used to evaluate Federal job training programs as well as other programs that are intended to increase employment and earnings. While the proposed WIOA regulations make significant progress in encouraging States to make better use of their own UI data for these purposes, using data compiled at the Federal level would improve our understanding of employment outcomes.  For example, it would streamline access to cross-State data (so that we can understand the employment outcomes for individuals who move across State lines after completing a program), and make it easier for the Departments of Labor and Education to conduct their own evaluations of WIOA programs without placing additional burdens on States.

The proposal would also allow the Census Bureau to access NDNH data for statistical uses, including for the 2020 Decennial Census.  Access to administrative data records, including the NDNH, could help the Census Bureau reduce the cost of the decennial census by $1.2 billion or more by using these records to identify who resides in non-responding households.  The proposal would prohibit the Federal statistical and evaluation units from releasing personally identifiable information, and it includes strong criminal penalties for individuals if they willfully make an unauthorized disclosure.

In addition, the Budget proposed expanding the scope of use of Medicare data by certain health research organizations and spur improvements in health care quality through activities like fraud prevention activities and value-added analysis for physicians to enable better care coordination and practice improvement. We are pleased that a similar provision was recently enacted in the Medicare Access and CHIP Reauthorization Act of 2015.

• Investing in data infrastructure. The Census Bureau has been a leader in working with administrative data. The Budget requests $10 million in additional funding for the Census Bureau to build on its existing strengths and start developing a more comprehensive infrastructure for linking, sharing, and analyzing key datasets. The additional funding would allow Census to accelerate the process of acquiring additional key datasets, including State-held data, such as for the Supplemental Nutrition Assistance Program (SNAP); expand and improve its infrastructure for processing and linking data; and improve its infrastructure for making data available to outside researchers, while protecting individual privacy and data confidentiality.

The Budget would also:

• Invest in the Department of Education's student aid data infrastructure,
• Help States make better use of their administrative data by supporting the improvement of their workforce and education data systems, and
• Support the linking of Bureau of Justice Statistics data with the Department of Justice’s grants management system data in order to shed light on the effects of Federal funding on recidivism and other justice system outcomes.

• Creating a commission to lay the groundwork for future improvements. The 2016 Budget includes a proposal to create a commission that would make recommendations for how to make additional administrative data available for evaluation and other statistical uses by Federal and outside researchers, what legislative changes are needed to facilitate such access, and how to expand access while ensuring data security and fully protecting privacy and confidentiality, as well as how to improve data quality. Last week Representative Paul Ryan (R–WI) and Senator Patty Murray (D–WA) re-introduced the Evidence-Based Policymaking Commission Act of 2015, a very similar proposal that creates a bipartisan commission focused on making better use of administrative data for fostering the types of learning described above.

The administrative data package outlined in the Presidents 2016 Budget fits into the Budget’s broader emphasis on tackling challenging but important reforms that are integral to making government work better. Harnessing the full potential of administrative data can improve transparency and support efforts to hold programs and service providers accountable; allow Federal agencies to adopt private sector best practices for using data analytics to improve performance and customer service; support ongoing innovation, experimentation, and evaluation to learn what works; and permit a greater understanding of the different needs of different groups and communities. We look forward to working with Congress to unlock the potential of administrative data while maintaining a strong framework of privacy, confidentiality, and data security.

Aviva Aron-Dine is the Acting Deputy Director of the Office of Management and Budget.