The first session of the workshop dealt with individual and household determinants of child food insecurity and hunger. It focused on vulnerable populations (disabled, minority, and homeless) and considered potential risk factors, such as health and disability (of parents and/ or children), poverty, economic shocks, and resource constraints. It also addressed the role of knowledge and behaviors, such as financial behaviors, resource management strategies, and nutrition knowledge. Susan Parish, Brandeis University, moderated the session. Craig Gundersen, an economist in the Department of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign, and executive director of the National Soybean Research Laboratory, was the principal speaker. The two discussants were Alisha Coleman-Jensen, a sociologist in the Food Assistance Branch, Economic Research Service (ERS), U.S. Department of Agriculture (USDA), followed by Sanders Korenman, an economist and professor at the School of Public Affairs at Baruch College, City University of New York.
STATEMENT OF CRAIG GUNDERSEN1
Gundersen began by providing a brief description of the Household Food Security Survey Module (HFSSM) that has been part of the Current Population Survey (CPS) since 2001 (see Chapter 2) and commented on
1Gundersen (2013b) prepared a commissioned paper for the workshop.
the extent of food insecurity based on current data. As he detailed in his background paper (Gundersen, 2013b), the extent of food insecurity is at an all-time high, with many demonstrated negative health consequences associated with food insecurity (see also Chapter 7).
Gundersen said that food insecurity is a function of economic factors, demographic factors, and participation in food assistance programs—the typical model is for households, and some models are longitudinal. He focused on economic and demographic factors, noting the impact of food assistance programs would be discussed later in the workshop (see Chapter 6). Rather than provide a comprehensive overview of extant research,2 he said he would focus on some of the findings on determinants from the first round of grants funded through the Research Program on Childhood Hunger, a program funded by the USDA Food and Nutrition Service (FNS) and managed by the University of Kentucky’s Center for Poverty Research (UKCPR).3 He said that this research used a wide variety of econometric techniques and datasets and posed distinct questions. Some of the main findings regarding the determinants indicate certain categories of children are more likely to be food insecure after controlling for other factors, which include children with:
• an incarcerated parent (Wallace and Cox, 2012);
• a parent who is an immigrant (Balistreri, 2012);
• complicated household structures (Balistreri, 2012);
• a parent with disabilities (Balistreri, 2012);
• changing residences (Jacknowitz and Morrissey, 2012); and
• declines in maternal or child health (Jacknowitz and Morrissey, 2012).
He also noted the following analyses that indicate correlates of food insecurity in all households, including households with children, using cross-sectional data, such as:
• lack of financial management skills (Gundersen and Garasky, 2012);
• an American Indian head of household (Gundersen, 2008);
• at high risk of homelessness (Gundersen et al., 2003);
• no child support (Garasky and Stewart, 2007);
2Gundersen indicated that controlling for other factors—households that are more likely to be food insecure include those with lower incomes; those headed by a single parent, a non-Hispanic black, a Hispanic, or someone with less education; those with more children; and those who do not own their home. He noted that these results are consistent with previous work on food insecurity across population types.
3Gundersen and James Ziliak are the principal investigators on this grant.
• a noncustodial father who does not visit regularly (Garasky and Stewart, 2007);
• lack of access to social capital (Martin et al., 2004);
• summertime (Nord and Romig, 2006); and
• a cigarette smoker in the home (Cutler-Triggs et al., 2008).
He noted that in studies using panel datasets, dynamic factors have been associated with being at higher risk of food insecurity, including negative income shocks, lack of assets, changes in household composition, and becoming unemployed (Gundersen and Gruber, 2001; Leete and Bania, 2010; Ribar and Hamrick, 2003).
Open Questions About the Determinants
Gundersen observed that based on the literature a great deal is known about determinants of food insecurity among households with children but little about mechanisms. In this context, he said, “mechanisms” refer to what it is about the determinant that is related to food insecurity. For each determinant, the question is “why does it matter?”
Coleman-Jensen and Nord (2013) described the effect of disability status on food insecurity, noting that it is not fully known why disability status matters, or what it is about having a disability that means markedly higher rates of food insecurity. Possible reasons may differ depending on who in the household has a disability and the type of disability, but Gundersen said understanding the mechanisms helps guide the appropriate policy interventions.
Similarly, he stressed the importance of understanding the reasons why immigration status and level of education have an impact on food security.
Gundersen noted a number of open questions about income and food insecurity. Depending on the year, many poor households are food secure. What are they doing differently than similar households who are food insecure? Do they have better financial management skills? Do they have more knowledge about how to get by on less? Is there underreporting of income? Conversely, about 10 percent of households with incomes above the poverty line are food insecure, despite seemingly having enough money to be food secure. Is the poverty line appropriately defined? With the recent economic downturn, is it because families have lost income but have fixed expenses? Often, food is an expense in which people can cut back if they cannot cut back on a mortgage payment. Might it also be lack of knowledge about how to get by on less money, different expectations about what constitutes a sufficient amount of food, or lack of access to food assistance programs for families with income above 185 percent of the poverty threshold?
He went on to ask why having a grandparent in the household is protective against food insecurity for the children in the household. For example, is it because it makes for a less expensive form of childcare, is easier to prepare meals at lower cost, or makes it easier to be eligible for Supplemental Nutrition Assistance Program (SNAP) benefits?
Finally, he asked, how do determinants differ by whether the household reports that they participated in food assistance programs? How do changes in household structure impact food insecurity, especially if there is a discontinuity in SNAP benefits? Does household size make a difference in determinants? How do determinants differ by whether the children are receiving meals through the National School Lunch Program or the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC)? He said it is well established that older children have higher rates of food insecurity than younger children and that rates of participation in the National School Lunch Program and WIC decline quite a bit as children age. Are those two things connected? Gundersen said that it has been shown that the National School Lunch Program is protective against food insecurity.
How to Interpret Determinants
Gundersen stated that more information is needed about the relative magnitudes of food insecurity determinants in order to more effectively target scarce resources. Gundersen went on to say that the food insecurity literature (like that on poverty) has tended to say that someone is either food insecure or food secure, but measures are also needed for the depth and severity of food insecurity. Clearly, he said, if someone responds affirmatively to 15 questions on the food insecurity module, that household is much worse off than the household of someone who responds affirmatively to 3 questions. He observed that in some earlier work (Gundersen, 2008), he described an approach to more fully utilize the 18 HFSSM questions to better understand incidence, depth, and severity of food insecurity. Some determinants might be correlated with the reporting of food insecurity. For example, people with higher education levels may be less likely to report that they are food insecure because they have a different understanding of the questions. Gunderson said further exploration of how people interpret and answer the questions should be considered. He also noted that the food insecurity literature in the United States and the food insecurity literature in developing countries have been relatively separate and distinct, and that many insights could be gained by cross-fertilization.
Gundersen asked whether consequences of food insecurity might actually be determinants of food insecurity. He noted sometimes it is clear that
one or another factor should be an exogenous determinant of food insecurity, but, in other cases, causality is not clear. An example is a mother’s depression. If a mother cannot feed her child, that might be depressive. However, if a mother has untreated depression, it may impair her ability to feed her child. Gundersen suggested employing econometric methods to more adequately determine the causal direction of food insecurity and its correlates.
Gundersen asked why data on food expenditures are often inconsistent with responses to the food insecurity questions. Along with the HFSSM, the December supplement of the CPS asks other questions regarding food-related topics. One of these questions is, “In order to buy just enough food to meet (your needs/the needs of your household), would you need to spend more than you do now, or could you spend less?” This is followed by a question, “About how much more (less) would you need to spend each week to buy just enough food to meet the needs of your household?” He suggested the combination of the responses to these questions and the HFSSM could provide researchers with a new way of interpreting the extent of a family’s food insecurity. Overall, he said, these questions have been underutilized in the study of food insecurity and have not been included on other surveys, although he has done some work on the question (e.g., Gundersen and Ribar, 2011; Gundersen et al., 2012a). Gundersen also wondered how the effects of determinants might differ if children rather than adults were asked the questions, a topic discussed later in the workshop (see Chapter 9).
Gundersen observed that while many of the research directions noted above can be pursued with existing datasets, some can only be pursued with new data collection efforts because most studies use nationally representative data and may exclude some groups, such as persons who are homeless or marginally housed. Better understanding of these overlooked groups will be important to policy responses because these groups are likely to have higher food insecurity rates than the general population.
Omission from the sampling frame (e.g., the homeless) is one reason why some people are overlooked. Another is survey nonresponse—higher levels of nonresponse may be associated with the marginally housed, immigrants, or individuals without immigration documentation. Nonresponse may contribute to bias in estimates for the overall population.
Gundersen went on to say that, to date, most understanding about food insecurity in the United States is based on quantitative datasets. In contrast, there has been very little recent work using qualitative data and, furthermore, the work that has been done has not had much influence on
the food insecurity literature or the policy ideas that have been generated. To give a more complete picture of food insecurity in the United States, he said more research with qualitative data would be worthwhile, both to address questions that quantitative data cannot address and to help inform development of new questions for quantitative data collection efforts. Gundersen offered three suggestions for qualitative research: (1) development of questions that quantitative data cannot address; (2) sampling of both food-secure and food-insecure households; and (3) use of transdisciplinary teams to allow for richer approaches to studying food insecurity. There is a need, Gundersen noted, for longer panel datasets over longer time periods to get at not only the dynamic determinants but also the duration of food insecurity. Examples of longitudinal datasets include the Panel Study of Income Dynamics and the National Longitudinal Survey of Youth–1979. In part due to the relatively recent development of the HFSSM, there are not any datasets with a long panel of observations on food insecurity. Datasets with multiple years of observations would allow for more effective use of econometric panel methods.
Gundersen concluded by saying that there is a need for more data with specifics about food spending and food access. He cited the National Food Acquisition and Purchase Survey, sponsored by ERS and fielded by Mathematica Policy Research in 2012, as a promising new survey and database for analysis.4
STATEMENT OF ALISHA COLEMAN-JENSEN
Coleman-Jensen stated that the main reason to understand determinants of children’s food insecurity data and the mechanisms by which they operate is to improve the design and targeting of programs and policies meant to improve food security. Important questions are where research efforts should be focused to address the determinants of food insecurity by level of severity and what the mechanisms are by which determinants affect food insecurity. Another question is whether it is more important to identify new determinants or to figure out how to use the information about known determinants.
Citing data from 2011 (as presented by Coleman-Jensen et al., 2012), she noted that the majority of households with children, 79.4 percent, were food secure. However, almost 21 percent were food insecure. Only the adults were food insecure in 10.6 percent of these households with children. In the remaining 10 percent, children also experienced the effects
4As of the date of this publication, the data were not yet available. Information about the survey is available at http://www.ers.usda.gov/topics/food-nutrition-assistance/supplemental-nutrition-assistance-program-(snap)/national-food-study.aspx [August 2, 2013].
of food insecurity with direct evidence of reductions in children’s dietary quality and quantity. One percent of households with children experienced the most severe food insecurity—the parents reported that the children were not getting enough to eat because of lack of resources for food. She noted further that from research on outcomes of food insecurity, children in food-insecure households have detrimental effects on their development—even if there is no evidence that the children themselves had reductions in their dietary quality or quantity. There may be more severe effects of food insecurity for households in which children actually are not getting enough to eat.
The Healthy, Hunger-Free Kids Act of 2010 (P.L. 111-296) calls for research on the causes of childhood hunger and food insecurity, which Coleman-Jensen said justifies focusing on any and all levels of severity. However, she noted, where to focus the investment remains an open question, pointing out that some recent UKCPR research focuses on the severe conditions of food insecurity among children with very low food security. Very low food security among children is difficult to study because it is relatively rare (as noted above, 1 percent of households with children). Even in large national datasets, the sample sizes are relatively small: The CPS is one of the largest datasets that include food insecurity questions, yet even in the CPS, which is the source for USDA statistics on food security, only 127 households in the sample had very low food security among children in 2011. This limits the research questions that can be addressed with these datasets. Coleman-Jensen suggested a need to learn more from specific efforts, such as the Witnesses to Hunger Project (see Chapter 5), focusing on the most vulnerable populations likely to experience severe conditions.
Current analysis assumes that the determinants of food insecurity, at the broader levels of the food insecurity, also affect very low food security. Coleman-Jensen stated that this is probably justified given that most parents will try to protect children from experiencing the more severe conditions. Investing in ways to help parents maintain their food security may also help children, she said.
She referred to research (Coleman-Jensen and Nord, 2013) that found that disabilities are an important risk factor for food insecurity. Not only was food insecurity more prevalent in these households, but also it tended to be more severe, with much more very low food security in these households than the researchers expected. She said more research is needed to identify how disabilities affect food security. She agreed with Gundersen about the need to understand the mechanisms of how disabilities (and other determinants) affect food security, especially if a policy goal is to reduce food insecurity.
Using the CPS data for 2010–2011, Coleman-Jensen noted among households with children, if an adult in the household is unable to work
due to disability, both the incidence rate and the severity rate of food insecurity among children are more than double those rates among households with children without a disabled adult. This indicates that disability is a very important risk factor for food insecurity.
Coleman-Jensen stated almost nothing is known about how disabilities among children affect food security. She referred to research by Parish et al. (2008), who examined material hardship among households raising children with disabilities and found higher food and other types of hardships. An important finding is that for households without disabled children, the number of hardships decreased as income increased above the poverty line, but this was not the case with households with disabled children. Coleman-Jensen noted her research with Nord also found households that include disabled members need a lot more income to make up for the costs associated with disabilities.
She pointed to new research opportunities in this area. FNS has funded data collection using the HFSSM 10-item adult food security questions in the National Health Interview Survey (NHIS), and this module was included in the 2011, 2012, and 2013 NHIS surveys. The NHIS includes a wealth of data on disabilities and health impairments for all household members and makes it possible to examine children and adults with disabilities in the same households. She characterized this dataset as a great research opportunity.
Coleman-Jensen noted that in their research Nord raised a question about how much variation in food insecurity is explained by known determinants, asking how much of the variation would be explained if all of the known determinants were contained in one study. She referred to a study by Bartfeld and Dunifon (2006) in which they explained variations in state-level hunger rates5 using a variety of household factors and state factors. They found about 86 percent of the variation was explained. However, she said it is not known whether this analysis would also apply to household food insecurity.
She returned to her earlier question about the need to decide whether to invest more in identifying new determinants or to invest more in better understanding already identified determinants. She said determinants are important, but so is translating those determinants into policy, perhaps targeting specific populations. In addition, she noted that it is important to balance identifying key determinants of food insecurity with understanding the characteristics of food-insecure households.
She displayed Figure 3-1, showing results from the CPS for 2010–2011 to illustrate the prevalence rate and severity rate of food insecurity among
5State-level “hunger rate” is the prevalence rate or the share of households that are very low food secure.
FIGURE 3-1 Prevalence and distribution of food insecurity among children, by employment and labor force status of adults in the household, 2010–2011 average from the Household Food Security Survey Module of the Current Population Survey (CPS-HFSSM).
SOURCE: Coleman-Jensen, McFall, and Nord (2013: Fig. 6). Reprinted with permission.
households in various situations. The graph shows that the majority of households with children (almost 60 percent) include a full-time worker and that prevalence of food insecurity is much lower for households where one parent or more are employed full time (about 7 percent) and much higher for households with unemployed adults (more than 25 percent) or in which adults are unable to work due to disability (more than 29 percent). Among food-insecure households with children, more than 15 percent have adults with only part-time work, more than 12 percent have adults who are unemployed and looking for work, more than 6 percent have disabled adults with no adult in the workforce, and 6 percent have no household member in the labor force for reasons other than disability.
Coleman-Jensen noted while it is important to target unemployed households and those with disabilities, the majority of food-insecure households with children would be missed if only those households were targeted. She said it is important to identify determinants and risk factors, but it is also important to keep in mind the entire population of food-insecure households.
She identified several potential research questions. In general, current food assistance programs target low income as the primary determinant of food insecurity. They all have income tests with the implied assumption that higher-income households do not need those programs. As Gundersen pointed out, this is not always the case. Is it possible to effectively target other determinants of food insecurity with policies or programs, Coleman-Jensen asked. Examples might include determinants such as time constraints around food preparation, lack of financial management skills, or physical disabilities that make it difficult to get to a store.
She asked whether specific programs should be targeted to specific populations, such as policies or programs to target persons with disabilities, or whether there should be less targeting and instead more general programs. For example, SNAP, available to low-income households, has special provisions for persons with disabilities. People with disabilities can deduct their medical expenses from their monthly income, which effectively raises their SNAP benefit. She said, however, that some research suggests that more needs to be done for households with disabilities.
She asked what level of severity should be targeted—the tip of the iceberg or the whole iceberg?6 The “tip of the iceberg” might represent households with food insecurity among children or, at the very tip, very low food security among children. The “whole iceberg” is the less severe condition that affects more children—food insecurity in households with children. Is it possible to shrink the tip of the iceberg without shrinking
6She attributed this analogy to Mark Nord.
the whole iceberg? Coleman-Jensen said she thinks the whole iceberg should be targeted, noting however, that this is an open area for discussion that can guide research investments.
Ultimately, she concluded, the question is how to invest the money from the Healthy, Hunger-Free Kids Act on research that will improve food security. Understanding the determinants and mechanisms is necessary to improve the design and targeting of programs and policies intended to improve food security.
STATEMENT OF SANDERS KORENMAN
The theme of Korenman’s remarks was that measurement matters, and measurement issues affect key variables that are central to a study about the determinants of food insecurity. He said the most important constructs are poverty, food insecurity, and program participation, especially participation in SNAP, and he praised research of Craig Gundersen, Mark Nord, David Ribar, and others on the topic. However, he posited, the understanding of measurement problems may yet not be deep enough to inform policy analysis concerning the determinants of food insecurity.
Korenman stated that priorities should be to improve measurement validity of the key determinant variables, understand reasons for measurement problems, and support research on determinants informed by an improved understanding of the key measurement problems. He said there has been some research on this, but not enough, and noted the goal of the research would be both to improve measures and to improve the estimation and interpretation of determinants in order to guide policy.
To explain why measurement matters for determinants, Korenman returned to Gundersen’s open questions about why so many poor households are food secure and why so many nonpoor households are food insecure. He noted measurement error could produce this pattern of results. He questioned whether the implication is that more work is needed to explain this result as a real phenomenon, or whether the result suggests the need for more valid measures of poverty to use in the analyses. Another question is how determinants differ by whether or not study subjects are participating in SNAP. The ability to answer is going to be affected by mismeasurement of SNAP participation, he said, especially if there is interest in determinants among poor SNAP participants and nonparticipants since mismeasurement of poverty compounds these errors. He referred to work by Meyer and Sullivan (2012a:117) that has shown that underreporters of income are disproportionately represented at the bottom of the income distribution, particularly by those whose incomes are below half of the poverty line.
He said he agreed with Gundersen’s suggestion of an analysis to
determine the relative magnitudes of various determinants, but mismeasurement would affect this analysis. In general, the more a determinant is mismeasured, the smaller the magnitude that will be estimated, which will influence the impression of which determinants matter more than others.
Korenman stated an analysis of how different determinants influence responses to food security questions is particularly important. He noted that there has been some analysis about how participation in SNAP affects responses to food security questions, but further research is needed. He described what he termed two mysteries. The first leads to concerns about the external validity of both the food expenditure measure and the food insecurity measure, as Gundersen and Ribar (2011) concluded that food hardships are underreported at the low end of the expenditure distribution. He cited a report by Nord (2009b) on the same topic that showed that even at essentially zero food expenditures, household food insecurity is still only at about 20 percent. Korenman stated that the data may be masking genuine distress, and it may mean that the food insecurity and insufficiency measures will have difficulty registering increases in well-being from policy innovations and economic improvement. The validity issues affecting these measures affect the ability to answer key questions of interest.
His second mystery, he said, is why SNAP participation is not inversely related to food insecurity among the poor. He referred to Gundersen and Kreider (2008), who observed that the apparent paradox of food stamp households appearing to be more likely to be food insecure than similar eligible nonparticipating households hinges on strong assumptions about the reliability of the data. They find that error rates in food stamp participation smaller than 12 percent would be sufficient to mask the relationship between food stamp participation and food insecurity. Korenman said that Gundersen and Kreider (2008) described measurement issues associated with the reporting of the food insecurity items, and food insecurity measures may reflect a different point in time than reported food stamp receipt. Korenman said that validity issues should become a more central part of research on determinates of food insecurity. He urged that poverty measurement and its validity be a concern for all studies concerning the determinants of food insecurity for several reasons. First, if poverty is measured poorly, it is going to bias estimates of the effects of poverty on food insecurity. Poverty is directly used as a partial screen in the HFSSM interview. It is also going to be important when looking at the determinants of food insecurity among the poor. He noted the poverty measure matters for important issues like identifying the needy and for assessment of policy effects.
He provided two brief examples. First, he noted that Meyer and
Sullivan (2012a) brought measurement issues to broad attention by comparing a consumption-based poverty measure to income-based measures of poverty like the official measure and the supplemental measure for determining the most disadvantaged, which Korenman termed a critical point. The identification of the most disadvantaged depends on how poverty is measured. As a second example of the fundamental importance of measurement and validity issues, Korenman considered whether the War on Poverty continued or failed to have sustained impact on poverty after the early 1970s. He said the answer to this question appears to depend on the poverty measure used. Discussing a graph from Meyer and Sullivan (2102b:149) that displays three poverty measures from 1960–2010, he noted that the official income poverty measure is relatively flat, showing little change after the early 1970s. The consumption poverty measure, available since 1984, shows much greater reductions in the poverty rate and thus implies a different story about the success of policy and the economy in reducing poverty over the past 30 years.7
He said Kaushal et al. (2012) took the issue of poverty measurement and brought it to bear on food insecurity, an example of the kind of measurement-focused research on food security that Korenman said he advocates. Kaushal et al. (2012) compared the magnitude of the relationship between food insecurity and poverty using the official poverty measure and the supplemental poverty measure. The supplemental poverty line or threshold in this case is greater than the official threshold, so each increment in income relative to that threshold is a bigger step up the income distribution. They found a bigger effect of increases in income relative to the poverty line when the higher supplemental poverty line is used. Thus, Korenman noted, a potential problem with the Kaushal et al. (2012) approach is that it may confound the rate of poverty with the measure itself. (To explore this, Korenman suggested that they could compare the effects of the two measures when they are “anchored” at the same rate.) Nonetheless, he stated that this paper illustrates important points and addresses the right questions.
Korenman speculated about a source of error that may or may not be unappreciated. He went through the order of the questions on the CPS questionnaire, noting the differences in the questions asked for poor and nonpoor households. He posited that respondents to the CPS questionnaire, particularly SNAP participants, may suspect they are being moni-
7During the open discussion, Joel Berg noted that between 1960 and 1974, the third measure shown on the graph (the after-tax money income measure) cut poverty in half, so that 16 million Americans left poverty and entered the middle class. (The official measure declined from about 19 percent to 11 percent.) Korenman added that since 1974 there has been little progress according to one measure and tremendous progress according to the other.
tored for compliance since, for example, questions about SNAP participation are asked before the HFSSM food security questions. Respondents may believe that agencies share data and may be concerned that their benefits might be jeopardized or reduced, thus affecting their responses. Many recipients are aware of stereotypes held by some people of the poor and of food stamp recipients. Korenman stated some interesting work could be done on measurement in this area, especially with the help of ethnographers. While investigators, including Gundersen, have been doing work on these measurement issues, he called for more research that puts measurement issues at the center of studies on the determinants of food insecurity.
Edward Frongillo (University of South Carolina) noted that both Gundersen and Coleman-Jensen asked how much of the variation in food insecurity is explained by hypothesized determinants, which determinants are the most important, and what the relative magnitudes of those determinants are. He said a different set of questions is who is food insecure, which families and which children are food insecure, and why. He posited that some of the methodology used is driving some of the answers that result, referring to research by Olson et al. (1997). Their classification and regression trees method showed that combinations of factors mattered in complex ways: for example, the variables of having some savings, food expenditures, whether solely reliant on food stamps for food expenditures, and whether there was some extra income for food worked well together. It is the combination of different factors that really matters. Gundersen said regression methods do not deal with combinations of factors very well, and other methods might be better.
Sharon Kirkpatrick (National Cancer Institute) commented on Gundersen’s question about why some low-income households are food secure and some higher-income households are food insecure by positing that income is an imperfect measure of actual household resources because it does not capture costs of housing and other basic needs, savings, and debt.
Deborah Frank (Boston University) suggested a reason why disability might contribute to food insecurity is that for children (and adults) with diabetes and food allergies, the costs of the recommended diet are much higher, but it is not clear how to capture this on national surveys. She also commented that the concept of “getting by”—which might mean nobody in the family experiences discomfort—is very different than everybody in the family getting enough of the right foods
for health. The Healthy Eating Index8 is complicated, and the linkage between “getting by” and the family’s Healthy Eating Index is a missing piece in the story, she suggested.
Korenman asked about research that looks at the relationship between health insurance and food security. Katherine Alaimo (Michigan State University) described research she conducted some time ago, noting that families who did not have health insurance were twice as likely to be food insufficient (Alaimo et al., 2001a). She went on to say that it is important to broaden the perspective about determinants of food insecurity and consider the fundamental causes of poverty in the United States. Some demographic characteristics are associated with poverty, but, she asked, why is it that poverty is more prevalent among African American and Hispanic households? The root causes, such as racism and the economic structure of U.S. society, are not usually discussed. She also noted that 80 percent of food-insecure households have members who are working, which indicates they are not earning enough. Alaimo commented that, in terms of financial management skills, it is hard to manage a budget when the amount of money is not enough. Whether it is important to try to teach people how to manage a very small amount of money is an open question. She said more qualitative research on coping skills is needed to be able to say that lack of financial management skills is a cause of food insecurity. Gundersen agreed it is important to increase people’s income, but he noted it might be beyond the scope of food insecurity research. He went on to say there is a debate in the literature about how much people can be taught financial management skills, with some evidence that instruction does not help much and that people seem to have those skills or not.
A participant referred to one of Gundersen’s slides, which stated having an older child was an important factor in the level of food security. It led the questioner to wonder whether work has been done to dissect the parents’ protective instinct versus the fact that as a child gets older, there is an increasing caloric need that puts the child at a level above the adults in the household. The participant queried whether there is a significantly different interaction for preschool children and younger children, how it changes over time, and whether it changes differently in different sub-populations. Gundersen agreed that this point may be one reason that older children have higher levels of food insecurity than those who are younger.
Joel Berg (New York City Coalition Against Hunger) commented on Korenman’s discussion about the official poverty rate and whether or not it indicates that the War on Poverty did not work. Even though the
official poverty rate was cut in half from 1960–1974, extending the line through the Reagan era and the Bush era is like saying that disease did not go away when penicillin was introduced or that penicillin was taken away and now the disease is at the same level. He called for separate discussion of the two time periods. He agreed that underreporting is a key issue when looking at the CPS and SNAP participation, noting that the CPS underreports SNAP participation by 30 percent. He also agreed with Korenman’s point about the questions asked as part of the CPS. A survey taker asks questions that a child welfare worker would ask, and a respondent may fear that answering the question negatively could mean losing his or her children, a much more severe consequence than losing benefits. He called for research comparing the United States to societies that have essentially eliminated hunger and food insecurity, citing the Scandinavian countries’ experiences. He suggested that if the preponderance of evidence shows the most meaningful factors affecting food insecurity are unemployment and poverty, more collective efforts should address income and poverty.