The Effects of SNAP on Family Interactions - Evidence from an Instrumental Variables Approach

Background: Studies on effects of social safety net program participation predominantly focus on economic outcomes (e.g., income, labor supply, family expenditure) and outcomes relevant to human capital development (e.g., education and health). The extent to which welfare programs may affect family dynamics has remained a neglected area of research. Understanding the effects of welfare program participation on family interactions can help contextualize and interpret mixed or weak findings on other outcomes and shed lights on the mechanisms in which programs exert their effects. This study investigates the effects of a major safety net program in the U.S., Supplemental Nutrition Assistance Program (SNAP), on family interactions. Theoretically, SNAP participation can improve family interactions through two pathways. First, SNAP participation increases family resources for acquiring food, which may free up income for enrichment activities for children and reduce financial stress faced by parents. Second, the increase in food intake may enhance health and wellbeing of both parents and children, which can further offer positive energy for improved family interactions. This study adopts a rigorous design to empirically examine whether SNAP participation can result in improved family interactions.

Data and Methods: This study uses data from the nationally-representative Survey on Income and Program Participation (1999-2011; 1996, 2001, 2004 and 2008 panels). The topical modules provide rich information on family interactions, including cognitive stimulation opportunities, family routines and life structures, parent-child interactions, family stress, and parental expectation. To address the problem of nonrandom selection into program participation, we exploit the variation in SNAP policies and administration across state and year and employ an instrumental variables (IV) approach to identify the causal effects of SNAP participation on family interactions. The key instrumental variables are biometric identification technology (e.g. fingerprinting scanning), broad-based categorical eligibility (BBCE), the re-certification periods, and outreach efforts. Our sample includes children aged 0 to 17 with a guardian who did not receive any college education and living in households at or below 130% of federal poverty level (N = 18,952). We control for a full set of household- and state-level controls and state and year fixed effects in the IV models. We conduct the analysis pooling all children together as well as separately by age groups (0-5, 6-11, and 12-17). The F-statistics for all four models (pooled and by age groups) range between 10.43 to 12.80, which are above the commonly acceptable cut-off of 10.

Results and Conclusion: SNAP participation among low-income families did not increase cognitive stimulation opportunities (reading time or extracurricular activity participation), family routines (family rules on watching TV or shared meal times), parent-child interactions (family outings and positive parenting in forms of praising the child or having fun time with them), and parental expectation of the child’s educational attainment. One exception is that among children aged 6-11, there is some evidence that SNAP participation statistically significantly reduced parental stress in caring for children.

Back to: ePoster Presentations I

<< Previous Abstract | Next Abstract >>