How automation reduced administrative burdens and increased access to Emergency Rental Assistance

New research shows how data can increase access and program integrity

From Don: There is so much bad news about governance that I love to highlight examples of when government innovations works. See for example, this piece about improvement in FAFSAs completion rates, or the use of automation to protect Medicaid coverage. This is another example of how governments used data to make life easier for citizens at a crucial life moment - when they are facing eviction.

The COVID-19 pandemic wrecked havoc on the lives of many low-income renters. Media headlines in the summer of 2021 warned of two impending challenges.

First: the expiration of the federal eviction moratorium in August 2021, which put many renters behind on rent at risk of eviction.

Second: despite the federal government earmarking over $46 billion in Emergency Rental Assistance (ERA), state and local governments struggled to get the funds to needy renters.

Outlets like The Washington Post and The New York Times reported some state and local governments were facing month-long delays in distributing aid as staff were burdened with an overwhelming number of “onerous” ERA applications, with “bureaucratic missteps” in the process.

Amidst these challenges, our recent research published at the Proceedings of the National Academy of Sciences (PNAS) shows a bright spot: some local governments used innovations to broaden access to rental assistance. Creative use of geographic data substantially increased access to rental assistance while lightening workloads for capacity-strapped local bureaucrats.

While this story unfolded with a specific program (Emergency Rental Assistance) during a particular setting (the COVID-19 pandemic), we think it offers broad lessons that could be applied to a range of social assistance efforts.

The innovation: using neighborhood estimates to supplement self-attested eligibility

One of the most complex rental assistance application requirements for both applicants and staff was verifying that renters fell below the program’s income thresholds. Many renters, lacking access to a computer, were applying on their phones and faced difficulty uploading PDF documentation; others had cash-based employment or unstable work histories.

In response to these challenges, the Treasury Department developed guidelines that enabled local governments to use an innovation called a fact-specific proxy (FSP). An FSP uses publicly available hyperlocal area-level income estimates from the US Census Bureau as a proxy for the tenant’s own income. FSP made qualified renters who signed a binding self-attestation and lived in areas with large shares of low-income residents automatically eligible to apply for ERA, without uploading additional documentation beyond this written attestation to verify their own incomes. It’s an example of how the government can use technology and administrative data to help rather than hurt people.

As the Treasury Department noted, the key ingredient of the FSP was the written attestation: “While a proxy provides reassurance about the probable truth of an applicant’s self-attestation, it is the applicant’s written attestation that in the end assures us of the truth of their stated income.”

FSP enabled housing agency staff to fast-track eligible applications from low-income areas with signed income attestations through a review process with far fewer administrative burdens placed on the applicant and far fewer demands placed on staff that were already overwhelmed. In other words, the innovation was a win-win, reducing burdens on both the clients and the government, rather than shifting frictions between the two parties.

How does it work? The Figure below shows how applicants who (1) self-attested as eligible and (2) resided in FSP zip codes faced fewer burdens to prove eligibility.

What was the impact?

A key challenge with studying innovations to address administrative burdens like the fact-specific proxy is how to establish causality: it would be infeasible and unethical to randomly assign some ERA applicants to face fewer administrative burdens than other applicants.

Our research takes advantage of idiosyncrasies in how the innovation was implemented to generate causal evidence for the effects of reducing administrative burdens. The paper contains more details of the design, but broadly, the proxy was implemented based on strict, zip-code level median household income thresholds. This means we can use observational causal inference techniques to compare outcomes for applicants who are otherwise very similar but who reside in zip codes that fall very close to either side of that statistically arbitrary threshold.

The figure below shows that implementing FSP increased the probability that an applicant was approved for rental assistance by approximately 15 percentage points. Strengthening our confidence in the results is the built-in replication: we find similar effects across two states that implemented the same innovation but in different ways, with different staff members, types of applicants, and other contextual factors.

These results show that two very different jurisdictions were able to implement a burden-reducing change that significantly increased access to help for low-income renters.

We also find that not only did the change increase access to help, it increased the speed at which renters received funds, an important outcome amidst the expiration of the eviction moratorium and mounting rental debt. The figure below, focused on Virginia where agency officials tracked precise application and funding timing, shows that FSP applications spent, on average, about two weeks less being processed than non-FSP applications.

How did FSP produce these changes? Discussions with program staff revealed that FSP not only made things easier for applicants by requiring them to identify and upload fewer complex documents, it also freed up significant staff time that had been devoted to back and forth with applicants trying to obtain valid documents. This underscores how reductions in administrative burdens not only help beneficiaries but can also help reallocate staff attention to the speedy and efficient distribution of funds.

What are the broader lessons?

First, the federal government can enable local burden reduction efforts: the federal government, often cast as an entity that stifles innovation in the safety net, set the conditions for local governments to adopt innovations like the proxy by publishing detailed guidance and spotlighting successful local case studies. This makes administrative burden targeting innovations less risky for local governments because they know they have federal backing.

Second, data creates opportunities: the use of publicly-available data to help target funds—in this case, American Community Survey (ACS) estimates of the income composition within zip codes—is a way to streamline access to support without needing to invest in complicated technological changes to application systems. Governments sometimes discourage and sometimes encourage the use of self-attestations. The use of FSP data provides more confidence in the credibility of those attestations.

Third, automation helps when bandwidth is limited: the results are especially relevant for situations like disaster relief where there is limited staff bandwidth and a premium on the timeliness of funds.

Tyler Simko is an Assistant Professor of Political Science at the University of Michigan. He specializes in US state and local politics, political geography, policy evaluation, and computational social science. He’s also a former local school board member.

Rebecca Johnson is an Assistant Professor at Georgetown’s McCourt School of Public Policy. She has a background in Demography and studies how social service agencies use a mix of data and discretion to allocate scarce resources.

Related stories on government innovations: