Files
Abstract
This paper examines how different econometric models, each with distinct functional form assumptions, influence estimates of the Supplemental Nutrition Assistance Program (SNAP) effect on food insecurity. To address potential endogeneity in SNAP participation, we employ a range of instrumental variables methods, including two-stage least squares, control function approaches, bivariate probit, and recursive bivariate probit models, analyzing SNAP’s impact under various linear and nonlinear specifications in both first-stage reduced form and second-stage structural equations. Using data from the 1996, 2001, and 2004 panels of the Survey of Income and Program Participation (SIPP), our findings reveal that the magnitude and direction of SNAP’s effect on food insecurity vary across model specifications, with estimates ranging from null effects to reductions of approximately 20 percentage points. Furthermore, heterogeneity analysis indicates that SNAP’s effectiveness is more pronounced among certain subpopulations.