UC Berkeley

This dissertation studies the role of information and psychological barriers to participation in three means-tested programs: emergency rental assistance (ERA), the Housing Choice Voucher (HCV) program, and the Supplemental Nutrition Assistance Program (SNAP). A growing literature on administrative burden documents three distinct costs that contribute to take-up gaps across the social safety net: learning (or information), compliance, and psychological costs. In parallel, a large body of empirical research focuses on designing and testing light-touch behaviorally-informed methods of reducing these barriers. Yet, these studies have yielded mixed results, suggesting that there is much left to understand about when, why, and for whom such methods work. This dissertation directly contributes to both of these literatures by extending the application of the administrative burden framework, examining all three dimensions of administrative burden, and testing behaviorally-informed methods of reducing information and psychological barriers.In Chapter 1, my co-author and I study the role of stigma as a barrier to take-up of emergency rental assistance. ERA programs aim to help low-income renters pay off back-owed rent and avoid eviction. In the wake of the COVID-19 pandemic, counties across the US received an unprecedented influx of federal funding for ERA programs. But many had trouble disbursing these funds to residents who needed assistance.

In two randomized experiments (N = 53,544; N = 62,528), we test whether reducing the stigma associated with ERA increases program take-up. In Austin, TX, we find that a destigmatizing email increases engagement by 36% relative to an information only email. We then build on these findings in a second field experiment in Denver, CO. In a mail-based outreach campaign, we find that a destigmatizing message significantly increases applications for rental assistance by 38% compared to a no-communication control group, and directionally by 11% relative to an information only communication. Moreover, we find suggestive evidence of larger effects for Black and Hispanic renters. In two subsequent online studies exploring mechanisms (total N = 1,258), we demonstrate that the destigmatizing language used in the outreach materials significantly reduces the internalized shame associated with participation in ERA, even in the presence of pervasive societal stigma.

In Chapter 2, I extend the administrative burden framework to consider the supply of services, rather than just demand for programs, and examine the barriers to landlord participation in the Housing Choice Voucher program in Minneapolis, MN through three methods: a large-scale survey, a randomized experiment, and a survey experiment. The HCV program provides very low-income individuals and families with vouchers (i.e., subsidies) to afford housing on the private rental market. This program is unique in that it requires participation both from residents and landlords. Landlords’ willingness to rent to tenants with vouchers directly determines both how many units are available, and where these units are located. Despite the centrality of landlords to the success of the program, there is relatively little empirical evidence on the barriers they face to participation, nor on effective methods for reducing them.

Existing evidence and narratives focus largely on the role of compliance hurdles—especially mandatory inspection requirements and time to lease-up—in deterring landlords from participating in the HCV program. Yet, in a survey conducted among all active landlords in Minneapolis (N ~ 15,000; response N = 1,088), I find evidence that psychological barriers, particularly the stigma associated with tenants who use vouchers, may be more consequential and pervasive than logistical barriers. In a field experiment (N = 13,419), I then test the impact of light-touch outreach that aimed to reduce these barriers to participation. Although outreach did not significantly affect landlord interest in the HCV program, in a subsequent survey experiment (N = 655), the same outreach materials had a directional impact on interest. This suggests the need for additional research to better understand the potential for light-touch strategies to increase landlord engagement.

In Chapter 3, I test the role of communication modality and message in increasing take-up of California’s SNAP program, CalFresh, among likely eligible college students (N = 275,977). Food insecurity among college students has increased significantly in recent years and is linked to a wide range of adverse health and education outcomes. However, just 20 to 30 percent of food insecure college students participate in the Supplemental Nutrition Assistance Program, in part because traditional eligibility criteria exclude most students. In response to the COVID-19 pandemic, SNAP eligibility was temporarily simplified and expanded. In turn, millions of low-income college students became newly eligible for benefits, which offered an opportunity to test the impact of targeted informational outreach on benefits take-up among a population that may face uniquely high barriers to participation.

In a large-scale randomized experiment, I find that simplified messaging increased application rates by 0.2 percentage points (pp), or 7%, relative to a status quo outreach message, but additional language tweaks aimed at reducing potential psychological costs associated with CalFresh participation had no impact beyond the simplified message. At the same time, I find that multimodal outreach (email and postcard) nearly doubled application rates compared to outreach conducted via email alone.