Overpayments of monthly benefits in Supplemental Nutrition Assistance Program (SNAP) vary widely across U.S. states—from $1.04 million in Wyoming to $865.4 million in California in fiscal year (FY) 2022—highlighting substantial differences in benefit delivery. This paper investigates the determinants of cross-state variation in SNAP overpayments per recipient from FY 2003 to FY 2019. Employing a LASSO variable selection framework, we assess the contribution of state economic conditions, caseload characteristics, and SNAP administrative policies—including simplified reporting, certification periods, and benefit calculation parameters. We find that the LASSO selected economic, SNAP caseload characteristics and SNAP policies explains 2.3%, 4.6% and 6.1% of the explained variation in overpayments, respectively, and state and year fixed effects account for 50% of the explained variation in overpayment of SNAP benefits. To explore the potential benefits of alternative administrative practices, we simulate FY 2019 overpayments under a counterfactual scenario in which all states adopt the characteristics of Idaho—the state with the lowest observed overpayment. The simulation suggests that total national overpayments could have been reduced by $1.4 billion (38%) relative to the observed $3.6 billion. These findings underscore the potential for reduction in administrative burden and targeted policy design to enhance SNAP payment accuracy rate.
