In an effort to expand participation and to ease the administrative burden of collecting and reviewing applications for the National School Lunch Program (NSLP) and the School Breakfast Program (SBP), the Healthy, Hunger-Free Kids Act of 2010 established the Community Eligibility Provision (CEP). Under the CEP, local educational agencies (LEAs) are given the option to switch to community eligibility, although not all schools within an LEA must participate. The provision requires that participating LEAs and schools serve both breakfast and lunch. As with direct certification, under the CEP students are categorically eligible on the basis of receipt of benefits from the Supplemental Nutrition Assistance Program, the Temporary Assistance for Needy Families Program, or Food Distribution Program on Indian Reservations, or because they are homeless, runaway, or migrant youth, or foster children. However, unlike direct certification, which confers eligibility to individual students (or families), the CEP allows schools, groups of schools, or entire districts to make free lunches and free breakfasts available to all students when 40 percent or more are directly certified as categorically eligible.
The CEP was initially implemented in a handful of pilot States, selected because they were thought to be most likely to have high numbers of eligible districts and schools. The program was implemented in Illinois, Kentucky, and Michigan in school year (SY) 2011-12; in the District of Columbia, New York, Ohio, and West Virginia in SY 2012-13; and in Florida, Georgia, Maryland, and Massachusetts in SY 2013-14, followed by nationwide implementation in SY 2014-15. Take up of the CEP has been substantial. Analyses by the Center on Budget and Policy Priorities and the Food Research & Action Center indicate that in SY 2015-16, 18,000 schools serving more than 8.5 million children had implemented the CEP, a take-up rate of greater than 50 percent, and 65 percent in the highest poverty schools.
The central research question guiding this report is whether attending a CEP school affects cognitive and behavioral outcomes for students. The study explores whether any predicted changes in child outcomes are more or less pronounced for children attending schools in rural areas. Because participation in rural areas has been lower, the changes in the administration of the NSLP and SBP driven by the CEP might differentially affect rural schools and thus child outcomes. In service of these questions, we use nationally representative panel data from the Early Childhood Longitudinal Study¬—¬Kindergarten Cohort 2010-11 (ECLS11).
The main independent variable of interest is a school-level measure of CEP participation, coded as equal to one if a school is participating in the CEP for the first time and zero otherwise. The study focuses on a number of key variables: Item Response Theory scale scores for reading, mathematics, and science; child scores on the Numbers Reversed subtest of the Woodcock-Johnson III, a measure of cognitive capacity; and teacher reports of child self-control, interpersonal skills, and internalizing and externalizing problem behaviors. In all analyses, the study controls for a number of child, family, and school-level variables.
To help address concerns about selection in estimating the effects of CEP participation, the study uses differences-in-differences models (DD), comparing student outcomes before and after CEP in schools that adopted the CEP to changes in student outcomes over the same period for schools that did not. Student outcomes are compared across four periods: (1) the spring of kindergarten to the spring of first grade, (2) the fall of first grade to the fall of second grade, (3) the spring of first grade to the spring of second grade, and (4) the spring of second grade to the spring of third grade. To account for low rates of participation in the CEP (only 0.2 percent of ECLS11 students attended a school that had newly adopted the CEP in the first period, 1.4 percent in the second, 1.3 percent in the third, and 0.2 percent in the fourth), the study pools data across the multiple periods. Finally, the study undertakes two supplementary analyses using a difference-in-difference-in-difference (DDD) framework. Using the first three periods when NSLP is measured, the study extends the DD analyses to investigate if CEP participation has differential effects for children living in rural areas or among children participating in the NSLP.
Initial models suggest that the CEP operated as expected: first-year, school-level participation in the CEP was associated with a 0.175 (p<.001) increase in the probability that a child ate a free or reduced-price lunch (in the waves when NSLP participation was measured). However, there was not a significant effect of CEP rollout on NSLP participation for children living in rural areas. Results for the main analyses indicate that CEP participation was not significantly associated with any child academic outcomes but was associated with a (p<.01) significant decrease in teacher-rated child self-control. Similarly, analyses from the periods when NSLP participation was measured in the ECLS indicate that there were no significant effects of CEP roll out on child outcomes for children participating in the NSLP. However, compared to children in rural areas in non-CEP schools, children who attended a school in the first year the CEP was implemented had significantly lower science scale scores (p<.05) and significantly lower levels of teacher-reported internalizing behaviors (p<.05).
National data on the CEP indicate swift and widespread take up of the provision. However, these national trends did not translate into high levels of participation in the ECLS11, despite the large and representative nature of this panel dataset. The national rollout of the CEP during the 2014-15 school year will almost definitely result in significant increases in participation rates in the ECLS11 as more schools in pilot States adopt the CEP and the remaining majority of States become eligible to participate. However, the relatively small proportion of ECLS11 children who attended a CEP school in the first few years of the program’s phase-in limited the statistical power, sophistication, and external validity of the analyses. The limited number of significant findings attached to the CEP rollout should be considered as preliminary and subject to replication once more data are available.