Panel Paper: The Health and Nutrition Effects of SNAP

Thursday, November 7, 2013 : 4:00 PM
DuPont (Westin Georgetown)

*Names in bold indicate Presenter

Marianne Bitler, University of California, Irvine
The goal of this paper is to assess the existing state of knowledge about whether SNAP improves health and nutrition outcomes, and if so, which ones and by how much. We will start by surveying existing work on the topic, drawing from different disciplines including nutrition science, public health, medicine, and the social sciences.

In an era of fiscal crisis, knowing whether SNAP has any significant causal effect on health is crucial for informing policy decisions and policy makers. In this review, we will pay particular attention to the challenges researchers face in overcoming selection bias and identifying causal effects of the program, and we will assess the literature through that lens.  The fundamental challenge in program evaluation in general and in assessing the impact of SNAP in particular is that participants are not selected at random from the population. Thus, comparisons of those who use SNAP and those who don’t---even conditional on observable characteristics—may not be apples to apples comparisons.  To the extent that those who choose to participate in SNAP are negatively selected—as one might expect---SNAP recipients are likely to be less healthy, and thus estimates of the effects of SNAP could be biased downward.

We will discuss and critically assess existing efforts in the literature to adjust for this possible selection.  Methods used in the existing literature range from relatively simple comparisons of SNAP participants and non-participants, to methods which adjust for selection on observables (either via matching or similar methods or by regression), to those relying on policy changes and differences in differences estimation or event studies, to instrumental variables and other approaches which deal with selection on unobservables. We will also discuss experimental approaches or use of bounding methods in this setting.

Substantively, we will consider the following questions.

1)         Does SNAP improve health and nutrition outcomes?  If so, which ones and by how much?

2)         Is there a health effectiveness gradient between children and adults? That is, is SNAP more protective of health for one group than the other? 

3)         Can SNAP improve birth outcomes? Is there useful evidence beyond work showing that the introduction of SNAP leads to improvements in infant outcomes?

4)         Does a full stomach (via use of SNAP) lead to better school performance?

In addition to reviewing existing work, we will descriptively examine associations between SNAP use and health and nutrition outcomes as described in questions (1) – (4). We will use data from the Continuous National Health and Nutrition Examination Survey for 1999-2008 and the National Health Interview Survey over a similar period to look at these associations in a multivariate framework.