Poster #37 - Social Safety Net Participation Among Resettled Refugees: Comparing Estimation Approaches

• In Event: Saturday Poster Luncheon

Abstract

Since the passing of the Refugee Act in 1980, over three million refugees have been admitted through the U.S. Refugee Admissions Program (DHS, 2024). For decades, programs supporting the early social and economic incorporation of refugees have received bipartisan support and been spared from broader anti-immigrant policy changes (e.g., PRWORA, Public Charge). For the first time, H.R.1 revoked refugees’ eligibility for several prominent safety net programs, creating gaps in access until they adjust to lawful permanent residency. To inform state and local policy response, there is a critical need to better understand refugees’ utilization of these supports. 

Few data sources allow for the identification of this specific population. Researchers have commonly employed Census imputation approaches, which seek to identify those likely admitted as refugees based on a series of factors, such as year of admission and country of origin (e.g., Capps et al., 2015, Evans & Fitzgerald, 2017, HHS, 2024). Research using these approaches have found initially high rates of participation in social safety net programs that decline to near general population rates over time. While this is generally touted as evidence of economic incorporation and upward mobility, other factors may be at play. First, these analyses have largely ignored the role of state policy variation (e.g., Medicaid expansion, broad-based categorical eligibility) and changes in sociopolitical factors across time (e.g., COVID-19, anti-immigrant sentiment). Evidence suggests that resettled refugees might be particularly sensitive or responsive to these shifts, such as changes in labor market conditions (Bollinger & Hagstrom, 2008). Second, in addition to individuals born outside of the United States having one of the highest Census nonresponse rates (O’Hare, 2018), there are additional concerns about the accuracy of these data for recent arrivals.

To investigate these trends, this study compares estimates of refugees’ receipt of SNAP, Medicaid, and TANF using three data sources: the American Community Survey (ACS), the Annual Survey of Refugees (ASR), and the Virginia Longitudinal Data System (VLDS). Single-year ACS files for 2011-2024 were accessed through IPUMS. Census imputation approaches proposed by Capps and colleagues (2015) were used to produce weighted proportions. Public use data from the ASR 2016-2023 surveys were accessed through ICSPR and include cohorts that arrived between 2011-2022. Data were pooled and individual weights were applied. And finally, data from the Virginia Department of Social Services (DSS) were accessed through the VLDS. Individuals were identified as a resettled refugee admitted between 2011-2024 using records from the Office of New Americans and were then matched with other DSS records on program enrollment. These data represent a near census for the state.

Amongst these data sources, there are important points of variation, including how time since admission is calculated, the question of participation is framed, and the program is defined. Preliminary results suggest noticeable differences in trends, which shape the conclusions drawn. The final analysis will include subgroup analyses by age, household composition, and geographic region and the discussion will highlight the policy implications.

Author

• Lindsay Dusard, University of Pennsylvania
  Presenting Author