Association for Public Policy Analysis & Management

Around 30 million students participate in the National School Lunch Program each school day. Approximately 70% of these students are approved to receive their lunch free of charged or at a reduced price. In order to maintain program integrity, USDA requires that the eligibility of a portion of participating households is verified through submission of additional income details. School districts notify, through a variety of the communication methods, households selected for verification. Verification response rates vary widely across the country, but on average 40% do not verify eligibility, many because they do not respond to initial letters requesting additional information. A 2004 USDA study found that many of the households that failed to respond to verification requests were, in fact, eligible for the benefits that were initially awarded. This study sought to test the impact of modified communication timing on the response to request for verification.

Program Change
In order to increase verification response rates among households who are income eligible, we partnered with the USDA’s Food and Nutrition Service (FNS) to modify the timing of the distribution of verification letters. Standard practice is to send verification letters on or after October 1st each year, which is after initial applications have been received and after children have begun receiving meal benefits. The timing of the letters was altered so that verification requests were sent out in weekly batches beginning as soon as applications were received for the 2016 school year, so that (i) letter recipients would more readily associate the verification request with their recent application for NSLP and (ii) the more spread-out pace of verifications reduced administrative burden and allowed district staff to pursue each case.

Methods
Working four district partners and a private vendor, we modified current software to allow school districts to select applications for verification in more frequent batches. In four districts the verification sample was split in half, so that half of the selected households were randomly assigned to receive requests shortly after submitting their application. The remaining households — the “holdout group” — received their letter after October 1st (business-as-usual). The four partnering districts had a total sample of 3,391 applicants randomly selected for verification.

Results
Introducing rolling verification had mixed results in encouraging verification responses among NSLP applicants. Controlling for date of application and initial eligibility determination, we found that overall verification response rates increased by 2.5 percentage points (p=0.11, 95% CI [-0.01, 0.06]) for “treated” applicants. When analyzing each district separately, response rates increased for applicants in three districts but decreased in the fourth. Differences between districts in the effect of changes in verification timing could be due to differences in the applicant pools across districts, although the factors that determine these differences are not known.