Can We Automate the Safety Net?

Lessons policymakers should learn from previous efforts

Each year, millions of Americans who qualify for public benefits fail to receive them.

Consider health insurance: in 2022, around 15 million people, the bulk of the uninsured population, went without health insurance coverage despite being eligible for free or subsidized coverage through either Medicaid or the Affordable Care Act (ACA) Marketplaces.

This same dynamic exists in other public benefit programs. In a given year, roughly 7 million eligible Americans don’t participate in the Supplemental Nutrition Assistance Program (SNAP) and another 5 million don’t claim the Earned Income Tax Credit (EITC).

These missed benefits add up. Forgone EITC benefits total approximately $7 billion each year. They also address vital needs. Missing out on SNAP benefits means that kids might go hungry. Whether one has health insurance coverage can be a matter of life and death.

So why do so many eligible people lose out on benefits?

One explanation has to do with how programs are administered. Most public benefit programs in the United States require people to opt-in. That means they have to learn about the program, determine whether they qualify, gather the necessary documentation, complete the application, and navigate whatever extra administrative burdens the agency requires, such as an in-person interview or employer verification.

At each step of this process, eligible people fall away. They miss deadlines, lose paperwork, become confused by complex rules, or simply lack the time or know-how to navigate a burdensome process while managing the daily challenges of poverty. What’s worse, means-tested programs require regular redeterminations in which recipients must prove their eligibility anew.

The result is a safety net that catches some but lets many slip through. The recently enacted One Big Beautiful Bill Act (OBBBA) only compounds this problem, adding myriad new administrative burdens to Medicaid, the ACA Marketplaces, and SNAP. Recent reporting suggests that state agencies are buckling under the weight of these new federal rules.

The Promise of Automatic Enrollment

“Automatic enrollment” is one solution that has attracted attention from policymakers, administrators, and advocates. Automatic enrollment refers to when the government identifies eligible individuals through administrative data and enrolls them without requiring an affirmative application on the individual’s part. As a result, the default goes from opt-in to opt-out.

Psychologists and behavioral economists have long observed that default rules—rules that determine how a person is treated if they don’t make an active choice—tend to have major effects on program enrollment. One recent study found that suspending health insurance auto-enrollment in Massachusetts caused new enrollments to decline by 33 percent, an effect that is an order of magnitude greater than the effects of softer “nudges” such as providing information and assistance and up to twice as large as the impact of Massachusetts’ mandate penalty.

Programs like Medicare Part B and Social Security, which come close to automatically enrolling everyone who is eligible, unsurprisingly boast take-up rates of nearly 100 percent.

Implementation Challenges

Yet if automatic enrollment is such a powerful intervention, then why hasn’t it been more widely adopted? And how can we implement automatic enrollment more broadly across the safety net?

In a new working paper, Automating the Safety Net, we try to answer these questions by conducting an in-depth comparative evaluation across numerous public benefit programs. We consider both the challenges involved in implementing automatic enrollment as well as how to surmount these challenges. Our analysis is informed by 28 semi-structured interviews with policy experts, technologists, and government officials who work on public benefit administration. We sought out people who have advocated for or helped to implement automatic enrollment, rather than its critics or beneficiaries. As a result, their views capture how practitioners see the terrain, and are not necessarily a verdict on whether automatic enrollment is worth pursuing.

The interviews point to four main challenges:

1. Decentralization. Many federally funded public benefit programs are run by states, territories, tribes, and, in some cases, counties. As a result, any improvement must often be made across multiple systems. One expert described county-level public benefit administration as “one of the biggest hurdles” to automating public benefits. They explained: “If you wanted to make a change to the [Medicaid] ex parte rate, you need to do it 50 times in California. You need to convince 50 different software vendors. . . . It’s just a much harder change management process.” These challenges are compounded by the divergent levels of capacity across states. “Each state has different degrees of sophistication within their program, their systems,” one expert explained.

2. Obtaining necessary data. Automation also requires that the government have the necessary data to determine eligibility and sometimes benefit levels. In theory, agencies can get some of these data from other government programs, but there are challenges to doing so. Perhaps the biggest hurdle is that eligibility rules differ among programs, making the data collected for one program unusable for another.

   Moreover, sometimes governments can’t access the necessary data or the data simply don’t exist. Some benefit applications require an in-person assessment. Others require frequent or detailed information that’s not automatically reported to any government agency, such as monthly or quarterly income data. As one expert succinctly put it, “You can’t automate things you don’t know. And usually there’s something you don’t know.” Governments could purchase relevant data held by private companies, but such efforts would have to grapple with what Luke Farrell recently described as the “means-testing industrial complex.”

3. Administrative capacity. Many of our interviewees also lamented public agencies’ limited capacity, especially at the state level, to build the necessary technology and data systems to improve benefit administration. One expert told us that in the states, “there’s significantly limited resources to undertake tech solutions, including time and human capital and money.” Another expert linked this to the need for more funding, noting that there’s “just not a ton of R&D and innovation funding that happens or trickles to [the] state level for this work.” Not only does this make it harder for government to create new benefit systems, but also it makes it harder to maintain and update them.

4. Fears and values. Beyond the technical and administrative challenges discussed above, automating public benefits requires political support. Most obviously, policymakers must actually view it as a problem that eligible people are not receiving benefits. One interviewee observed that “There’s a massive political challenge: like half of our system thinks that… nobody should have Medicaid ever.”

   Automating enrollment also entails potential risks, both to the public and to enrollees themselves. For example, policymakers have long invoked concerns about “waste, fraud, and abuse” to justify imposing administrative burdens on public benefit programs. More recently, Congressional Republicans repeatedly invoked such concerns to justify enacting the OBBBA, even though that law, apart from the undoubtedly significant SNAP cost-sharing change, contains only a few minor provisions focused on fraud and abuse and though it will mostly serve to disenroll people who are in fact eligible.

   Several interviewees also worried that data shared with the aim of automating enrollment could be used by Immigration and Customs Enforcement (ICE). The potential upsides of administrative data sharing to help people may be checked because of a fear that it will be used in punitive ways.

Four Recommendations to Ease Safety Net Automation

While daunting, these challenges are not insurmountable. As we describe in our article, the United States has, in recent years, taken some promising steps toward automating enrollment across a range of multiple government benefit programs, including the Child Tax Credit, Medicaid, and Summer EBT.

In 2021, the IRS used prior-year tax returns to automatically deliver monthly Child Tax Credit payments to tens of millions of families, helping drive child poverty to a historic low of 5.2%. During the post-pandemic Medicaid unwinding, many states dramatically improved ex parte renewal processes, which kept millions of Americans covered without requiring them to reapply. And the Summer EBT program, made permanent in 2023, was designed from the ground up to use existing SNAP and school-meals data to enroll eligible children without requiring a separate application.

Interviewees suggested several steps legislators and regulators could take to address the challenges described above:

1. Aligning Eligibility Rules. Many experts noted that aligning definitions and eligibility rules across programs would facilitate automatic enrollment. Some also suggested that lawmakers make greater use of adjunctive eligibility, where qualifying for one program automatically qualifies someone for another.

   At least one expert recommended increasing the use of “area-based” eligibility that deems someone eligible based on where they live. One example of this type of approach is the community-eligibility option in the school meals program, wherein a school that serves a large percentage of low-income students can choose to provide free meals to all students without requiring applications.

2. Improving Data Access and Quality. Experts also emphasized the importance of improving agencies’ access to the kind of high-quality data necessary to automate public benefits. Some indicated that new legislation would be necessary to encourage or require agencies to share data with each other. One expert observed that it helps to “be very explicit in new legislation [about] how data can be used.” Another suggested that federal agencies should publish a standard data sharing agreement.

   To fill gaps in existing data, some experts suggested imposing more frequent income reporting requirements. As one expert put it, the government should “require employers to report income data to the federal government at every pay period.” Another expert likewise emphasized the importance of “building more in-house federal data sources” so that states are not as reliant on private companies such as Equifax to provide data to determine whether enrollees are eligible.

3. Increasing State Capacity. Many of our interviewees emphasized the importance of increasing states’ administrative and technical capacity. When asked what steps could facilitate automatic enrollment, one expert put it bluntly: “Increase state capacity for IT projects.” Another expert explained that “there needs to be a legislative-level investment in actually fixing the underlying technical infrastructure of all this stuff.”

   Several experts stressed, in particular, the value of hiring in-house engineers to help run benefit programs. One expert pointed to several states that had recently built up their technological expertise at their agencies, showing it can be done.

4. Building Trust and Facing Fears. One obvious way to build trust is to mitigate, as much as possible, the real risks automatic enrollment poses. Experts offered a few suggestions for how to do this. One expert emphasized the importance of “adopting a significant harm-mitigation strategy” that includes contingencies for when enrollees are improperly enrolled in benefit programs. Another expert recommended limiting the information being shared for automatic enrollment initiatives to “what’s necessary for enrolling someone or establishing that they’re eligible.” For instance, rather than sharing reams of detailed personal data, “you could just have a flag that says [this person is eligible], and all you need to transfer is the flag.”

   But good governance alone isn’t enough to build trust. One problem is that people often don’t notice the benefits government provides. Several experts therefore stressed the importance of communication, both to ensure enrollees know they have coverage and also to build public support and embolden other officials to follow suit.

   Experts also emphasized the importance of political or agency leadership in empowering and motivating civil servants to automate enrollment. For example, one expert recalled how a state official initially declined to implement a program making it easier for people to file state taxes, only to change their mind two weeks later after receiving a call from the Governor asking them to do so.

Automating Enrollment in an Age of Retrenchment

The OBBBA’s impacts are already being felt across the country and millions of Americans must navigate the new administrative burdens it imposes on Medicaid, the ACA exchanges, and SNAP. In addition, the OBBBA’s cost-sharing requirements in SNAP and its restrictions on Medicaid financing create new fiscal challenges for states that are causing them to further restrict access to public benefits programs.

As a result, this may seem like a bad time to consider ambitious strategies to expand access to public benefits. Yet politics can change quickly. When it does, we should be ready. Our law has made promises to people in need. Automatic enrollment may well be our best way to keep them.

Gabriel Scheffler is a Professor of Law at the University of Miami School of Law.

Ariel Jurow Kleiman is a Professor of Law at the USC Gould School of Law.

Andrew Hammond is a Professor of Law and Harry T. Ice Faculty Fellow at the Indiana University Maurer School of Law.

For more on automatic enrollment