James Ziliak is the founding director of the Center for Poverty Research, and Carol Martin is the Gatton Endowed Chair in Microeconomics in the Department of Economics at the University of Kentucky. Ziliak was the chair of the steering committee that developed the program for the workshop, and provided closing remarks, followed by open discussion.
Ziliak stated his purpose was to pull together recurring themes. He clarified that he was speaking for himself and not on behalf of the National Academies or the other members of the steering committee.
Ziliak described the charge to the steering committee and the key objectives of the workshop, based on the Healthy, Hunger-Free Kids Act of 2010 (see Chapter 1). He suggested that there are several fundamental questions. First, how adequate is the current state of knowledge? Second, do substantial knowledge gaps remain? Third, do data exist to support research to fill any remaining substantial knowledge gaps, and if not can such data been generated? Fourth, how great are the research opportunities in this area? It was with these questions in mind that Ziliak and others on the steering committee put together the workshop.
Ziliak said that in the area of identifying the determinants of hunger, the workshop covered much of the existing knowledge and issues, including the geographic and environmental factors associated with the risk of childhood hunger, the role of the social safety net (especially federal food
assistance programs), and the extent to which there are gaps in program coverage or inability to adequately access those programs. He said that one fundamental question posed by Allard (see Chapter 4) is outstanding: how is adequate access defined? This question remains as a challenge to be answered.
Regarding the consequences of childhood hunger, he noted the workshop participants discussed the public health and medical consequences in some depth. Much attention was focused on measurement and its implications for how child development, well-being, and food insecurity are understood. Ziliak asked about the adequacy of the current state of knowledge, noting tremendous progress over the last 15 years since the food security module was introduced into the Current Population Survey (CPS). The collection of these data has been a tremendous resource for research to move forward. He said that researchers have learned a lot about what is affecting low-income and disadvantaged populations in this country through the food insecurity measures and its links to child development and other outcomes. Even though the data have some issues and gaps, Ziliak said he believes that important research on child hunger has progressed because of the available data. Its strengths and importance can be acknowledged, along with a discussion about opportunities for improvement.
Ziliak noted recurrent themes expressed by several speakers concerning the knowledge gaps that remain. He said that, as others observed during the workshop, measurement matters, both how food insecurity is conceived and how the questions are asked. He said that Korenman raised a provocative issue about the screener, the sequence and issues that are associated with this screen, and the questions asked in the implementation of the food security supplement (see Chapter 3). Ziliak said that he and several colleagues have been examining food insecurity as measured in the Panel Study of Income Dynamics (PSID), using data collected in 1999, 2001, and 2003. The rates of food insecurity measured in that survey are about half of those measured on the CPS, and the question is why. He said that he and his colleagues will review measurement issues based on Korenman’s comments.
Ziliak said workshop presenters provided valuable information about the scale to measure food insecurity. He noted a recurring debate about whether narrowly measuring food insecurity or including marginal food insecurity is most important, which the advocacy community has also taken up. Ziliak reported that the Meals on Wheels Association of America and the National Foundation to End Senior Hunger, for which he and Gundersen have done work, prefer the marginal food insecurity measure, whereas Feeding America prefers the food insecurity measure. The best measure, he noted, could be different for different purposes. For
example, Meals on Wheels focuses on seniors, and Feeding America worries about the entire age distribution, while other advocacy groups focus on children. He suggested that the research community might provide better guidance to the advocacy community as to which measure seems to capture which types of phenomenon and which ones seem to be the most pressing.
Another theme discussed throughout the workshop, he noted, is the shortage of longitudinal data. He referred to the debate about whether or not causal pathways can be identified with cross-sectional data. Ziliak said he agrees with Gundersen that there are some abilities to address causal pathways with cross-sectional data, but many important questions can better be addressed with longitudinal data, such as issues of duration, transitions on and off, and the evolution of health status over time. The health profession, and health economists in particular, has created a literature about the gradient of health status, not only with the life cycle but also across the income spectrum. There is an interaction between age and income over the life cycle, and longitudinal data allow for a look at those interactions. He said that Chilton presented provocative ideas about multigeneration links between hardships in childhood and food insecurity.
Ziliak described the currently available data to address some of these longitudinal issues. He noted the Early Childhood Longitudinal Study-Birth (ECLS-B) and -Kindergarten (ECLS-K) Cohorts support answers to some limited questions but those data are limited to specific cohorts of children. The National Health and Nutrition Examination Survey is not a panel, and the CPS allows for the matching of at most one-half of the sample for two years. He said that one could do repeated cohorts, but that requires more assumptions than a true panel. The PSID is a true panel, and he said he has used the PSID data for 25 years,1 which he said has come to life for a new generation of research. The dataset follows families since 1968, their children, and their grandchildren. Starting in 1999, the PSID began collecting food security information as well as comprehensive information on health, consumption, and assets. With the PSID, a researcher has data from a true panel, with food security information as well as additional depth of information on consumption, health, and wealth.
The information on food insecurity is only available in three waves (1999, 2001, and 2003). The PSID is fielded every two years. A 10-year-old in 1999 will be 26 in 2015, and he called this an opportunity to add the food security scale back to the PSID to allow researchers to do analysis of
1Ziliak disclosed that he is on the Board of Overseers of the PSID, which is an unpaid position.
first intergenerational transmission of food security. He said researchers could look at the consequences of being food insecure as a child, as a young adult, and as a 25- to 35-year-old in 2015 (as suggested by Chilton in Chapter 5). That is just one example of potential research.
He said that the PSID has a comparatively small sample and statistical power is an ongoing challenge if the intent is to drill down to the concept of very low food insecurity. The only dataset Ziliak said he is aware of with an adequate sample size, offered on an ongoing basis, is the American Community Survey (ACS). The ACS funding is continually challenged in Congress, and he said the Census Bureau is careful about adding more items. He suggested that there may be an opportunity to propose the addition of the two-item screener discussed in Chapter 9 rather than the full 18 items, saying that if policy makers could find the financial will to add a food insecurity screener to the ACS (covering three million U.S. persons annually), the issue of statistical power for many research purposes would be addressed.
The analysis of existing data may illuminate important data items that do not exist today. Identifying and collecting new data is an area where demonstration projects could be useful. Ziliak said workshop funders may like to understand how demonstration projects could help to assess upcoming proposals and identify areas where further research is warranted. Ziliak noted themes expressed by some workshop participants included the issue of geographic differences in the cost of living, and he asked whether a demonstration project could examine variable Supplemental Nutrition Assistance Program (SNAP) benefits based on this factor. He referred to Shore-Sheppard’s work, described in Chapter 7, which includes a labor-intensive nonexperimental assessment of whether an extra dollar of SNAP benefits affects food insecurity. He also noted that the pilot study of the Summer Electronic Benefits Transfer for Children documented in Collins et al. (2012) indicates that additional SNAP benefits matter. Ziliak suggested this area might result in other excellent demonstration projects.
He said that population groups are important to study, such as immigrant families and the challenges and hardships they face. Another important population group, he said, is individuals with physical or mental disabilities, whether the parent in the household, the child, or both.
Ziliak acknowledged tremendous momentum on the research spectrum of childhood hunger, and suggested keeping that momentum moving, perhaps through a system of centers of excellence. Ziliak noted that the assistant secretary for planning and evaluation in the U.S. Department of Health and Human Services has been running poverty research centers since the 1960s, providing one existing model, and the population research centers of the National Institutes of Child Health and Human
Development provide another example. The population and poverty centers are slightly different from the Research Innovation and Development Grants in Economics Program in that the former have a very strong component of mentoring the next generations of scholars. Ziliak noted Diane Schanzenbach (see Chapter 8) mentioned the importance of funding graduate students in her presentation, which he said he thinks is a fundamental component of a center.
Ziliak suggested creating a network of scholars, bringing together the human capital that attended this workshop, and conducting workshops on a more regular basis to keep a community of scholars connected with one another—bringing together ideas from the multiple disciplines of medicine, public health, economics, sociology, social work, nutrition, and others.
Ziliak made two additional suggestions. He said he and Gundersen have evaluated many proposals in the last several years, and one challenge is that not all the surveys use the 18-item scale. If they use a shortened scale, different surveys may use different subsets of questions. One suggestion would be to establish funds to create a questionnaire bank. Researchers could apply to use those questions.
Second, he said a funding mechanism that would allow new questions to be added to existing surveys would be important. He suggested that if researchers identify a common set of questions and develop a mechanism to help fund that subset on different surveys, even a competitive program might provide another mechanism to keep up the momentum and branch out to important new sources of data.
David Smallwood (Economic Research Service) stated that he was speaking as a program manager and funder of research, rather than a researcher. From that perspective, he stressed the importance of more data and models with a big emphasis on data. Research centers are great, he said, but data are essential for research to move forward. He stated that, for example, most of the research presented at this workshop was based on data that the U.S. Department of Agriculture (USDA) had the foresight to fund years ago, including the measure in the CPS, PSID, and ECLS-B and ECLS-K. In looking to the future, new kinds of data will be necessary to foster new insights. Otherwise, he asserted, researchers will be faced with diminishing returns from current data, which are themselves quite limited for studying very low food insecurity in children. For example, he referred to the fact that only 127 households with very low levels of food insecurity of children were identified in the most recent year of the CPS. Are researchers going to do case studies on those? The CPS data
cost about $600,000 a year to collect, which he said is very reasonable because it is an add-on module to the CPS survey. He noted that if longitudinal data are important, USDA could add the food security module to an additional survey. However, to find enough observations to examine very low food security over time in such a survey might take more money than USDA has available. He suggested thinking outside the box on data, especially using existing monitoring systems, such as the school system monitoring approach that Frongillo mentioned. He noted certain biological markers could inform different data collections. He referred back to the modeling idea and the iceberg metaphor that Coleman-Jensen used (Chapter 3): Rather than just looking at the 127 households in the CPS at the tip of the iceberg, researchers could use data from all households to model the entire iceberg and then extrapolate from the base to the tip.
Ziliak responded by encouraging greater linking of administrative data to currently existing data. He noted that in looking at SNAP effects on food insecurity through survey data, there are selection problems or reverse causation. There is also measurement error from underreporting. The linking of administrative data to existing datasets can solve some of those problems, although good mechanics are needed to deal with reverse causation.
Mandy Murphy (University of California, Berkeley) commented that the workshop illustrated the problem of food insecurity among children, but researchers seem to be unclear about how to best define the problem. If the community tries to completely and perfectly understand and define the problem by improving measurements, methodology, and understanding the determinants, she said, the problem probably will have shifted by the time it is defined. Instead, she invited and challenged participants to focus on innovative solutions by including the people to be helped as an integral part of the conversation and ultimately the solution. Using the community resources in place, creating new solutions, and looking at how to improve the federal assistance food programs would be ideal, she said. Ziliak noted at a time of financial and fiscal retraction, evidence-based policy development is key. He reiterated the need to work toward creative solutions to the problem.
Jay Hirschman (Food and Nutrition Service) thanked the speakers and the National Academies for the workshop. He concurred that it will be important to keep the network alive and interacting, recognizing the unique opportunity of research funding provided under the Healthy, Hunger-Free Kids Act of 2010, Section 141. He also called attention to language in Section 141 that authorizes demonstration projects with truly rigorous evaluation to provide the knowledge and results of long-term lasting value. He noted that USDA wants to ensure that research and demonstration projects are coupled to provide the best, well-planned ideas.