On February 2, Abt Associates and the Association for Public Policy Analysis & Management (APPAM)
are jointly hosting a forum
on rethinking food assistance policy in the United States. The forum will highlight evidence — in particular, two recent Abt Associates studies — with relevance for ongoing discussions of the effectiveness of food assistance in promoting food security and nutrition.
Food Assistance Programs in the U.S
To address food insecurity in the United States, the U.S. Department of Agriculture manages 15 food assistance programs designed to provide children and low-income people access to food and a healthful diet. The largest of these programs, the Supplemental Nutrition Assistance Program (SNAP), serves one in seven Americans and cost nearly $70 billion in 2015.
Despite its size — and many efforts over the years to evaluate its impacts — important questions about SNAP’s effectiveness persist. Does SNAP and programs like it improve food security? Do the benefits contribute to a healthy diet? While, on the face of it, it seems logical that giving people food assistance will increase their food expenditures and the quality of food purchased, conventional economic theory suggests that impacts may be small. Low-income households, who face a limited budget and competing demands for their resources, might substitute the food assistance for the cash they used to spend on food and then use the cash for other purposes. In that case, food assistance might add to household resources but not solve the hunger problem. Furthermore, even to the extent that households do use food assistance to purchase more food, they might purchase food with low nutritional value.
Using Random Assignment to Detect Impacts of Food Assistance Programs
Whether food assistance increases the quantity or quality of food for low-income households can be informed by empirical evidence, the best of which comes from evaluations with random assignment designs. In these designs, the only systematic difference between those chosen and not chosen for an intervention is by chance — a flip of the coin, so to speak. One can then say with confidence that the intervention accounts for any differences in results between the two groups. Random assignment evaluations of SNAP have not been possible as it is an entitlement so cutting benefits randomly is illegal. Increasing the benefits for a randomly selected group would require funds that have not been available.
Food Assistance Reduces Food Insecurity
Two studies led by Abt Associates used random assignment to shed light on the questions of whether food assistance programs can improve food security and diet quality. The first is an evaluation of the Summer Electronic Benefit Transfer (EBT) for Children Demonstration (SEBTC), conducted with Mathematica Policy Research and MAXIMUS. SEBTC was part of an $86 million initiative that provided summer food assistance to children who received free and reduced-price breakfasts and lunches during the school year. The study showed that the summer food assistance –$60 per month per child – cut the most severe form of food insecurity among children by one third. Further analysis by Abt helps answer the broader question of the impact of additional food assistance on food security of households with children. We estimate that $100 of additional food assistance per household per month would decrease household food insecurity by 30 percent.
Food Assistance Policies Can Shape Food Choices
While there is mounting evidence that food assistance can improve food security, a related question asks whether food assistance programs help improve nutrition. Evidence from the Summer EBT for Children evaluation also sheds light on the question of whether food assistance can improve the nutrition of school-age children.
Grantees who received food assistance funds chose to use either a SNAP or WIC EBT systems to deliver SEBTC benefits. WIC – the Special Supplemental Nutrition Program for Women, Infants and Children—provides nutritional assistance to women, infants, and pre-school age children as well as other services.
Those who used the WIC model provided randomly selected households with credit to obtain a specific set of foods appropriate for school-age children, selected from existing components of the standard WIC food package. Households receiving SEBTC through the SNAP system could use the benefits for all SNAP-eligible foods. The evaluation showed that while both the SNAP and WIC models improved children’s nutrition, the WIC model achieved greater improvements.
A second random assignment evaluation, the Healthy Incentives Pilot (HIP) led by Abt in partnership with Westat, also provides insights on how food assistance policies can shape the food choices of recipients. In 2011-2012, the study randomly assigned SNAP households to receive a 30 cent rebate on every dollar of SNAP benefits spent on qualifying fruits and vegetables with SNAP benefits. HIP households spent $6.15 – about 9 percent – more each month on fruits and vegetables than non-HIP households. Survey results show that HIP participants ate about a quarter of a cup more qualifying fruits and vegetables each day, eliminating a fifth of the total fruit and vegetable deficit relative to the dietary guidelines. Thus, a substantial financial incentive can moderately improve dietary intake.
During the APPAM/Abt forum, researchers and policymakers will gather to further discuss findings from these two evaluations — considered in the context of other important research evidence — and their policy implications. Abt hopes that the two evaluations and resulting dialogue will inform the further development of the nation’s food assistance programs.
Read more about this work:
Read more about Ann Collins.
It is hard for most of us to imagine going to bed hungry or worried that we can’t feed our family because we lack money to do so. Unfortunately 15 percent of American households—48 million people—face this situation. For these people, food security—defined as having access to food at all times and enough food for an active, healthy life — is elusive.