Files

Abstract

Empirical methods for estimating the treatment effects of the Supplemental Nutrition Assistance Program (SNAP) routinely focus on the average treatment effect of the program. This statistic is satisfactory and useful for many policy makers, although researchers understand that it is unlikely that program effects are constant across the treatment population. Obviously, differences in treatment across observed household, individual or geographic characteristics could lead to heterogeneous outcomes. And there are good reasons to think that effects of treatment will vary across \textit{unobserved} factors in household: food preferences, subjective poverty thresholds, discount rates, and financial acumen all could affect the distribution of outcomes not captured in the mean treatment effect. We estimate finite mixture models in order to address heterogeneity in response to receipt of SNAP and find that the data suggests two latent classes of recipients: one for whom SNAP has little or no effect, and one for whom SNAP has large and significant effects. This is true for both of the outcomes that we examine: food spending and food insecurity.

Details

PDF

Statistics

from
to
Export
Download Full History