Association for Public Policy Analysis & Management

The Supplemental Nutrition Assistance Program (SNAP), the National School Lunch Program (NSLP), and the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) are the three largest US Department of Agriculture nutrition assistance programs. While all three are primarily federally funded, their administration is complex at the state and local level. All three programs are overseen by state departments of public health, education, and social services, but are implemented by local agencies (county welfare departments, school districts, and local WIC agencies). These programs have overlapping goals – in particular, ensuring adequate nutrition among children from birth through school age. The goal of program integration across SNAP, NSLP, and WIC can be seen diverse ways, for example the categorical eligibility of SNAP recipients for NSLP, targets for directly certifying students on SNAP for NSLP, and a provision under discussion in the reauthorization of the Child Nutrition Act to extend eligibility for WIC to children until they are age-eligible for kindergarten. However, the realization of that policy aim is complicated by the complexity of the governing relationships. Cohesive estimates of the current extent of program integration are also lacking. This paper will provide these estimates, focusing on the two-way combinations of SNAP and NSLP and SNAP and WIC, but also assessing the extent to which children who age out of WIC eligibility have not yet “aged in” to NSLP eligibility. We focus on California, relying on the augmented American Community Survey (ACS) datasets for 2011-2013 prepared to compute SPM-style poverty estimates for California. These so-called California Poverty Measure (CPM) estimates lean heavily on accurate reporting of participation in social safety net programs (Bohn, Danielson, Levin, Mattingly, and Wimer, 2013; Wimer, Mattingly, Kimberlin, Fisher, Danielson, and Bohn, 2015). The ACS has relatively limited reporting of means-tested programs, but a very large sample, making it suitable for robust state-level and sub-state level estimates. Given known, substantial underreporting of program participation and limitations in the ACS questionnaire, the need to impute multiple program participation in the ACS is defensible. Nonetheless, care in imputing programs jointly is critical. The paper will detail our approach to imputing participation in SNAP, WIC, and NSLP and will provide upper and lower bounds on the joint probabilities of participation given known marginal probabilities taken from detailed California administrative data (via a partnership with the California Department of Social Services). The resulting estimates can help policymakers assess the coverage of these programs both statewide and regionally, informing state and federal efforts to link the programs more closely together.