Older Adults and SNAP: Using Linked Administrative Data to Understand Program Participation and Duration

Background

An aging American population puts upward pressure on the safety net. The Supplemental Nutrition Assistance Program (SNAP) is the nation’s primary food assistance program targeted to needy families, serving over 42 million people (Cunnyngham, 2018). In the absence of SNAP benefits, 3.2 million people would experience poverty (Fox, 2019). Yet, SNAP enrollment for older adults lags behind that of other age groups (Vigil, 2019). There are likely sub groups of older adults who participate for long and short spells, yet little is known about these patterns. This study uniquely contributes to the literature by describing differences in SNAP participation and program composition between older and younger adults.

Method

We assembled a unique data set from state administrative records comprised of all Oregonian individuals who received SNAP and/or Temporary Aid for Needy Families (TANF) at any point in the five year period from 2014 to 2018. The data included program participation (e.g., benefits receipt and amount) and demographic characteristics (age, sex, race/ethnicity, and geographic location). The resulting five-year panel included 7,807,640 person-year records across five years. This study focused on long term spells on SNAP defined as 5 sequential years of benefits receipt. We measured SNAP participation as having received SNAP benefits at any time in the calendar year. Multivariate regression with clustered standard errors at the person were used to predict factors associated with short- and long-term participation patterns. We also used survey data from the American Community Survey (ACS), accessed via IPUMS (Ruggles, et al, 2019) and augmented with TRIM3 microdata (Urban Institute, 2014), to assess overall Oregon SNAP participation. Survey-weighted state-representative descriptive statistics on SNAP participation were calculated.

Results

ACS data showed that overall SNAP participation declined slightly from 26 to 24.9 percent of the total Oregon population. Older adults (ages 55+) were about half as likely to participate in SNAP (15%) compared to younger groups (29%). These patterns were mostly stable over time. However, the administrative data reveal that older adult participants were much more likely to participate continuously for five years (43% older / 27% younger). After controlling for other demographics, geographic location, and time, the probability of being a continuous SNAP participant was highest (49.6%) for those aged 70 to 79 compared to those under 50 (26.7%). A within-group regression of older adults revealed that women, African Americans, and metro residents had relatively high risk of sequential SNAP participation.

Discussion

Using a combination of administrative data from safety net program participation along with survey data, we reveal a new policy problem: older adults in Oregon are both less likely to participate in SNAP overall, but more likely to experience long spells when they do participate. Findings are constrained to the period under examination and limited by left- and right-censoring. Nevertheless, findings present difficult challenges for policymakers and practitioners who seek to ensure access for older adults in need. The full paper links income and earnings histories and considers more dynamic measures of program participation.