Identifying the Effect of WIC on Infant Health When Participation is Endogenous and Misreported

The existing evaluations of the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) agree on a beneficial association with birth weight but not necessarily gestation age. Regardless, considerable doubt exists over whether these associations represent a causal relationship. Endogenous selection into WIC, lack of valid exclusion restrictions, and rampant under-reporting of participation are to blame. Here, I utilize the nonparametric bounds method in Kreider et al. (2011) to address both identification problems simultaneously to assess the causal effect of prenatal WIC participation on birth outcomes. In addition, I complement the partial identification approach by reporting instrumental variable estimates following Lewbel (2010) to circumvent the need for a traditional instrument. Using data from the ECLS-B, I show that ignoring misreporting and only accounting for self-selection, WIC improves birth weight and, sometimes, gestation age. However, if only one percent of eligible women misreport their participation, well below the expected level of misreporting, the effect of WIC on birth outcomes cannot be signed.