Association for Public Policy Analysis & Management

Food insecurity has emerged as a key public health challenge in the United States since the Great Recession. This emergence has been especially strong due to (a) the significant increase in and persistence of high food insecurity levels since 2010 and (b) the extensive negative health and other outcomes associated with food insecurity. In response, an extensive literature on food insecurity has been developed which has greatly expanded our understanding of the causes and consequences of food insecurity.

This literature has, in virtually all cases, concentrated on a binary measure of food insecurity, namely, whether or not someone lives in a food insecure household. This binary measure is based on an underlying set of 18 questions on the Core Food Security Module (CFSM) whereby someone is classified as food insecure if there are three or more affirmative responses. As a consequence, if someone responds affirmatively to five questions, they are treated identically as someone who responds affirmatively to 15 questions. The dichotomous treatment of the responses obscures the fact that the latter person is reporting far greater distress than the former, an important consideration for policymakers and practitioners wishing to construct effective interventions.

In this paper, we explore the utility of applying more detailed analyses of trend data with respect to measures of the depth and severity of food insecurity. Using methods based on the Foster-Greer-Thorbecke (FGT) class of poverty measures, Gundersen (2008) established a set of food insecurity measures that portray the incidence, depth (how far below the food insecurity cutoff a household is), and severity (applying greater weight to those farther below the food insecurity cutoff) of food insecurity. Using data from the Current Population Survey (CPS; the data set used to establish the official rates of food insecurity in the U.S.), we apply these methods to examine what has happened to food insecurity over the time period of 2002 to 2013. We consider this for both (a) all households and (b) households with children.

We find that the patterns differ quite substantially depending on which measure is used. For example, the standard incidence measure shows that food insecurity increased dramatically during the Great Recession and has not subsidized since then. In contrast, the depth and severity measures fell in 2007 and it wasn’t until 2010 that these measures increased to above the 2006 level. However, they have increased substantially since then, unlike the incidence measures which have remained steady since 2008. Thus, the shift in food insecurity from 2002 to 2013 is far higher under the depth and severity measures than under the incidence measure. We conclude with policy implications emerging from this work and provide some thoughts on how these measures can be better utilized in a broad array of food insecurity research.