A new report from USDA’s Economic Research Service says federal surveys can overestimate SNAP’s impact on food security by up to 50% compared to administrative records. ERS set out to show that administrative data offer a more accurate picture of how SNAP affects food security than the survey data typically used in federal research.
Historically, studies evaluating the Supplemental Nutrition Assistance Program have relied on the Bureau of Labor Statistics' Current Population Survey Food Security Supplement, but ERS found high rates of misreporting when comparing those responses with administrative records from 19 states between 2017 and 2019.
Nearly half of SNAP participants in the administrative data — 47.3% — did not report receiving benefits in the survey, while only 2% of survey respondents reported receiving SNAP despite not appearing in the administrative files.
ERS says these reporting errors distort SNAP’s measured impacts in opposite ways.
Survey reporting errors can stem from social stigma, survey design, memory lapses and respondent cooperativeness. These measurement issues have “led researchers to underestimate its [SNAP’s] effectiveness in reducing poverty.”
Additionally, SNAP households identified through administrative data appear more food secure than those identified through surveys alone. ERS found that “SNAP has a potentially large and positive effect on food security,” but administrative data produce more precise estimates than survey responses.
“These results affirm that SNAP impacts food security in a positive way, but they suggest that causal estimates of food security made with survey data alone may significantly overstate the ATE [average treatment effect] of participation,” the report said.
