Cost of Safety Net Research Sprint
Nationwide
June 2025 - September 2025
Why This Matters for Families
Implementation of the One Big Beautiful Bill Act (H.R. 1), signed into law in July of 2025, means that the size and shape of America’s “safety net” is changing – again. Tens of millions of people in America rely on safety net programs for health insurance (Medicaid), food (Supplemental Nutrition Assistance Program (SNAP)), and other forms of income assistance (such as the Women and Infant Children Benefit (WIC), Temporary Assistance to Needy Families (TANF), and Unemployment Insurance (UI)).
While H.R.1’s changes were mostly focused on Medicaid and SNAP, it opens up a broader question about not just the cost of administration to implement these changes, but how these two foundational programs are part of a broader, complex web of programs that critically serve people across the country, often in their most vulnerable moments and periods of life. Changes in Medicaid and SNAP will surely generate rippling impacts on the total resources available to serve families, access, and ultimately, outcomes we experience in communities.
As a starting point to understand these impacts more holistically, our “good government” field needs a clearer understanding of administrative costs overall, and how to know what constitutes a “good” level of administrative spending.
Implementation Challenge
H.R. 1 significantly changed eligibility for multiple safety net programs and shifted the balance of cost between federal and state governments. States have faced a wave of tight timelines to implement major changes to policy, staffing, business processes, IT systems, and developing supporting budget forecasts. Safety net programs are already an integrated nest of program design, administrative processes, federal appropriations, agency regulations, and reporting requirements. Moreover, our federalist system layers in yet another set of silos across the 50 states, the District of Columbia, and sometimes counties, territories, and tribal nations that implement safety net programs for residents.
For states or other stakeholders, tracking the basics of how much is being spent, by whom, and for what purpose across programs becomes a dizzying project. Nonetheless, understanding these factors is foundational to understanding how to achieve cost-effective implementation, strong performance and, ultimately, good outcomes for families that participate in these programs.
Our Approach
This sprint took an early, exploratory step toward a research agenda for the field that, though technical, can be instrumental to informing how families are served. Our goal is to help anchor a needed dialogue around cost-effective administration of the safety net and how to center robust, sustainable outcomes for families – especially if a window opens to transform and improve how the system works writ large.
Our team designed an eight-week research sprint to structure an approach towards measuring “the cost to implement” America’s safety net. With limited time, our project could not explore these challenges in depth or produce final answers. Instead, we hoped to identify and refine key questions, map available data, identify measures that might help advance the field, and conduct some initial analysis to identify bright spots of where there may be lessons to learn from leading states .
We began by asking questions that consider both cost efficiency and program effectiveness. Through a combination of desk research and interviews with program experts, we narrowed our scope to address the Supplemental Nutrition Assistance Program (SNAP), Temporary Assistance to Needy Families (TANF), Medicaid, and Unemployment Insurance (UI). We then conducted independent analysis and consulted with other researchers and policy experts on how to present our preliminary findings in a meaningful, digestible way that can connect to other projects across the field.
OBJECTIVES
We named starting questions to help understand what cost-effective administration might be and explored how well existing data might provide answers:
What is the “cost of the safety net” - e.g., primarily the total federal administrative dollars spent on each of these programs’ administration? If you add in state dollars, what is the total cost? Is it possible to tell how much of this cost is duplicative - e.g., the cost of reviewing the same basic eligibility information differently for different programs
Might we build an “efficiency ration”? - e.g., for every $1 of federal administrative cost, how many dollars are ultimately received by or paid for the direct benefit of an individual?
Do any states have particularly “good” administrative cost to benefits paid ratios? How do we know what a good ratio is?
Of those states that have “good” cost-to-benefit ratios, are the programs actually “working” (producing positive outcomes)? What should “working” mean, based on available data from payment error rates, to fraud rates, to coverage rates, to actual outcome measures like food insecurity and health measures?
Do form and process integration (“reducing burden”) reduce operational cost measures AND increase “experience” measures like timeliness of processing and time it actually takes to complete a form?
WHAT WE DID
Assessed the quality and completeness of potential data sources on cost and program performance across SNAP, TANF, Medicaid, and UI.
Conducted expert interviews with current and former federal leaders and budget analysts, academic and think tank researchers, and advocates to validate our key questions and vet data sources.
Compiled 10 years of retrospective spending data at the national level for initial analysis.
Selected the most recent two years of consistently available cost data for all four programs to analyze at the state level.
Conducted light statistical analysis to compare spending on administration versus benefits across states and to identify potential connections with program performance, state demographics, and political factors.
Shared high-level findings with researchers and state leaders for feedback.
What We Learned
Our project demonstrated that America’s safety net programs lack consistent, reliable, high-quality data about spending and program outcomes, quality, and burden. We consider the findings below to be preliminary “indicator lights” that signal a potential truth and call for further exploration.
Spending on administration versus benefits varies wildly across states as well as programs. No state has a consistently low ratio of spending on administration versus benefits across safety net programs.
No clear factors seemed to predict whether a state spends more or less on safety net administration – neither political control nor demographic factors like population size or density, racial and ethnic composition, or household income.
Program access and payment accuracy are sometimes in tension – and sometimes not. While these priorities appeared to be tradeoffs in SNAP, more states appeared able to perform well on both measures of access and of fraud in UI.
Integrating application forms is a time-saving win for applicants but, so far, isn’t clearly translating into internal efficiency gains for states.
These high-level findings raised numerous next questions for future projects and collaborations across the field. They also suggest that the field may be able to help states to identify successful policies and practices within individual programs that could be applied more broadly, such as to lessen tradeoffs that states make between providing access to programs and ensuring payment accuracy.
The passage and implementation of H.R.1 has ushered in an unusual period of major policy change for key safety net programs and potential fiscal impacts for states. Understanding the administrative choices, trade-offs, and best practices involved in delivering programs is critical to delivering a system that works for families. Moving forward, especially in a chaotic data environment, a structured approach to tracking the full “cost” of implementing H.R.1 – both in material administration and longer-term impacts – is necessary for informed policy conversations. As a next step, our team began a “Cost of H.R.1” research sprint to support states.
