Food Stamps and Food Insecurity: What Can BeLearned in the Presence of Non-classical MeasurementError?

Policymakers have been puzzled to observe that food stamp households appear more likely tobe food insecure than observationally similar eligible nonparticipating households. We reexamine this issueallowing for nonclassical reporting errors in food stamp participation and food insecurity. Extending theliterature on partially identi…ed parameters, we introduce a nonparametric framework that makes transparentwhat can be known about conditional probabilities when a binary outcome and conditioning variable areboth subject to nonclassical measurement error. We …nd that the food insecurity paradox hinges on strongassumptions about the reliability of the data that are not supported by the previous food stamp participationliterature.[...]