![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Friedman School of Nutrition Science and Policy, Tufts University, Boston, MA 02111
* To whom correspondence should be addressed. E-mail: parke.wilde{at}tufts.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT IntroductionLITERATURE CITED |
|---|
|
Introduction |
|---|
|
TOPABSTRACTIntroductionLITERATURE CITED |
|---|
FSP. In 2005, the FSP spent $31 billion and served a monthly average of 25.7 million low-income Americans. To be eligible, most households must have monthly gross income <130% of the federal poverty line ($2097 for a family of 4), monthly net income after certain deductions less than the poverty line ($1613 for a family of 4), and countable assets below a specified value (7). Although the program is a federal entitlement for many low-income Americans, others are categorically ineligible, including full-time college students, people on strike from work, certain classes of legal immigrants, all undocumented immigrants, able-bodied adults without children who have exhausted a 3-mo time limit, and former drug felons.
If the household is very poor, having a net cash income of 0, it receives the maximum benefit amount approximately equal to the monthly cost of the Thrifty Food Plan ($506 for a family of 4 in 2005). Otherwise, the household's food stamp benefit is reduced by 24 cents for each dollar of earnings and by 30 cents for each dollar of other cash income (7).
A thorough 2004 USDA literature review summarized the large body of research showing that food stamp benefits substantially raise food spending, may raise nutrient availability in household food supplies, and cannot be shown to consistently affect individual nutrient intake (8). At the time that review was written, the literature measuring the effect of food stamps on food insecurity and hunger was sparse.
Food insecurity and hunger measurement. A working group including the USDA and other federal agencies adopted new formal definitions of household food insecurity and food insecurity with hunger in the early 1990s based on earlier definitions suggested by the AIN (1,2). "Food security" is defined as "access by all people at all times to enough food for an active, healthy life" (9). "Food insecurity" is defined as the absence of food security at the household level. Hunger is defined as "the uneasy or painful sensation caused by a lack of food" and in practice measured as a severe stage of food insecurity, although the official usage of the term hunger may change in the near future (2). In 1995, the federal government began to measure the prevalence of these conditions each year using an 18-item battery of questions on a supplement to the Current Population Survey and other nationally representative household surveys.
Optimism about food insecurity and hunger measurement for program assessment. One goal of this effort was to measure progress toward national objectives. For example, the federal government's Healthy People 2010 objectives included a goal of reducing the national prevalence of food insecurity by one-half, to 6% of households (3). Another goal, expressed eloquently in early accounts of the new measurement project (1), was to assess and improve food assistance programs. Following a requirement that all federal agencies define program goals and measure progress toward specific performance targets, the USDA's Food and Nutrition Service (FNS) established an official target for reducing the prevalence of food insecurity with hunger among households with income below 130% of the federal poverty line (10), which is the gross income limit for the FSP. From a baseline of 11.8% in 1998, the strategic plan set a target for this prevalence of 7.9% by 2005 (Fig. 1). In a recent performance rating, the Office of Management and Budget certified the FSP as "moderately effective" and yet demanded more rigorous empirical evidence of the program's impact on nutrition outcomes (11). In response, FNS promised to develop studies to measure the impact of food stamp participation on dietary quality and food insecurity.
|
Only 60% of eligible people choose to participate (14), and those who suffer from hunger are more likely to take the trouble to participate. As a consequence, even if one restricts attention to the population of households with income below 130% of the poverty line, the prevalence of food insecurity with hunger is about twice as high among food stamp participants as among nonparticipant households (Fig. 1). This self-selection or self-targeting pattern has been noted many times in the recent literature (8,9,15,16). Efforts to address this problem with more complex statistical approaches have generated a series of interesting papers and articles that shed light on the self-selection pattern but do not in the end succeed in quantifying the effect of food stamps on food insecurity and hunger. This section reviews 7 such research approaches.
The first approach is to control for other observable variables while seeking to measure the effect of FSP participation in a regression model. However, several studies have found that prevalences of food insecurity or hunger remain much higher for participants than for nonparticipants even after including control variables in this fashion (17,18).
A second approach is to jointly model the effect of food stamps on food insecurity and vice versa using a system of simultaneous equations. This statistical methodology requires either a strong assumption about the distribution of the error terms or the presence of instrumental variables, which strongly affect FSP participation but do not otherwise affect food insecurity. Using cross-sectional data from the Survey of Income and Program Participation, Gundersen and Oliveira (17) reported that the more complex statistical models removed the evidence of a significant positive association between food stamp participation and food insufficiency. More precisely, their simultaneous equations models found as large a positive association between program participation and food insufficiency as simpler models did, but the standard error was larger in the more complex models and hence the significance disappeared (17; Table 3). Also using simultaneous equations, Jensen (19) found that the random disturbances for equations describing program participation and food security status were correlated, such that a higher tendency toward FSP participation was associated with lower risk of food insecurity. Using data from the Survey of Program Dynamics, Huffman and Jensen found that being food insecure with hunger strongly increased the likelihood of FSP participation but "no evidence that the food assistance reduces food insecurity" (20).
A third approach is to use longitudinal or panel data. Hofferth (21) used data from the Panel Study of Income Dynamics to measure the association between transitions in FSP participation status and food insecurity, but she concluded that most of the observed associations reflected variation in families' unmet needs for food rather than a true effect of program participation. Using data from the Survey of Income and Program Participation and Survey of Program Dynamics, Ribar and Hamrick found that food stamp use in 1994–5 was associated with lower rates of exit from food insecurity by 1997, and hence their study "provides no evidence that food stamps alleviate food problems," but they also noted the continued presence of unobservable factors and did not claim to have measured the causal impact of FSP participation (22). With longitudinal data from the same Current Population Survey that provides federal statistics on food insecurity and hunger, Wilde and Nord (23) found that using a fixed effects panel data model to control for time-invariant unobservable factors reduced but did not eliminate the appearance that FSP participation was associated with poorer food security status.
A fourth approach, pursued by Gibson-Davis and Foster (18) using data from the Early Childhood Longitudinal Survey, is propensity score matching. This statistical technique estimates a model for the probability of program participation then predicts the probability of participation for each person in the sample and finally compares the food security status of matched participants and nonparticipants with similar predicted probabilities. The authors' most reliable models indicated that food stamps did not affect the probability of being classified as food insecure, although they offered "tentative evidence" that food stamps reduced the level of food insecurity among those who are categorized as food insecure. The authors warned that establishing cause and effect with propensity score matching "might not be possible in some instances, given the magnitude of the selection effects."
A fifth approach, the "dose-response" approach, employs variation across participant households in the food stamp benefit amount to measure the response of food security status to an increasing dose of food stamps (8). Some early research using this approach found a small but significant association between higher benefit amounts and lower risk of food insecurity (24). Other work using a dose-response approach is ongoing (25,26). Kabbani and Kmeid (27) used a dose-response approach to study whether a household that experienced hunger in the previous year nevertheless reported food secure status in the most recent month. Interpreting results from the dose-response approach requires careful attention to the nature of the program's benefit formula, which implies that the poorest households generally receive the highest food stamp benefits. The source of variation in food stamp benefits conditional on cash income is not random, but rather it depends systematically on the household's deductions (28). As a consequence, it is difficult to distinguish the effect of a higher food stamp dose from the effect of lower cash income and differences in the household characteristics that influence deductions.
A sixth approach exploits "natural experiments," such as changes in program rules for some populations but not others, for use in a quasi-experimental research design. In the only such study we could find that directly addresses food stamps and food security, Bartfeld and Dunifon (29) used hierarchical linear modeling to investigate the effect of multiple variables including state-level FSP participation rates measured as a fraction of the state-level low-income population. A model including interaction terms between the state-level FSP participation rate and the household-level income status found small but significant beneficial effects of higher FSP participation rates for near-poor and low-income households but not for poor households or higher-income households. In a study of a related natural experiment, Borjas (30) found that immigrant populations whose program eligibility was restricted in the 1996 welfare reforms experienced a significant relative deterioration in food security status. Two studies have used variation in weather, a truly "natural" independent variable if ever there was one, to assess the effect of heating and cooling costs on food security status (31,32).
The seventh and final approach would use a random-assignment research design to measure the effect of FSP participation on food security status. Influential studies in San Diego and Alabama in the early 1990s used a random-assignment design to measure the effect of cashing out food stamps on food spending (8). Random-assignment research designs have been much discussed in the recent literature on food insecurity and hunger but never implemented. Several authors use identical language to hold up random-assignment designs as the "gold standard" but then go on to describe them as unethical or infeasible for the current research question (8,16,18).
In addition to the literature relating food stamps directly to food security, reviewed here, there is also a growing body of research relating food security status to weight status (33–35) and relating FSP participation to weight status (36–38). These studies draw on a conceptual understanding of the likely contribution of food stamp benefits to reduced risk of household food insecurity but do not seek to measure the effect of food stamps on food security status.
In conclusion, more than 10 y of concerted research have tempered the initial optimism about using the federal government's new survey tool for program assessment. FNS, its contractor Abt Associates Inc., and an outside panel of experts have struggled to develop a research design that will satisfy the agency's earlier commitment to the Office of Management and Budget. The contractor's recent report labeled the nonexperimental approach "extremely difficult, as illustrated by a long history of inconclusive research and the expert panel's rejection of the designs initially proposed in this project" (16). As a consequence of this difficulty, this research field stands at a crossroads, facing a stark choice between retreat in one direction and stronger research designs in the other.
A recent high-profile report from the Committee on National Statistics urged the USDA to consider alternative terminology to describe severe conditions of food insecurity, because the word "hunger" may imply a more serious condition than is actually indicated by the current category "food insecure with hunger," while validated tools for measuring an individual-level concept of "hunger" have not yet been created. The report articulated the case for lowered expectations in measuring policy impacts, arguing that even an appropriate food insecurity or hunger measure "would not be a definitive performance indicator of food assistance programs" (2). In the meantime, it has become clear that the target in the FNS strategic plan will not be met, and indeed that the prevalence of food insecurity with hunger has risen in the low-income population (Fig. 1). Whereas the USDA strategic plan for 2002–7 specified a clear quantitative target for the prevalence of food insecurity with hunger among low-income Americans (4), the most recent USDA strategic plan for 2005–10 omits such a target, saying that it will be determined following review of the recommendations from the Committee on National Statistics (39).
If, instead of giving up on program evaluation, stronger research designs are pursued, there are several feasible and ethical approaches to consider. One of the studies proposed by Abt Associates Inc. would use a random-assignment research design to compare diet quality and food security outcomes for eligible FSP nonparticipants who either received or did not receive special encouragement to sign up for benefits (16). Many other promising research ideas share the key features of that proposal, including satisfying the ethical requirement that all subjects receive at least the "current standard of care" and that no subjects receive less than the benefits to which they are entitled under federal law. For example, a participant/nonparticipant comparison could provide benefits to randomly selected poor people who have an ineligible immigration status, who are former drug felons, who are childless and have exceeded their 3-mo time limit, or who own a disqualifying asset; a dose-response analysis could provide an additional food stamp benefit to randomly assigned households whose low deduction amount happened to qualify them for an atypically low food stamp benefit amount; a cash-out study could provide flexible benefits in place of targeted food stamp benefits, to measure the distinct effect of food stamps on food security status while holding constant total household resources; and participants could be assigned at random to have their benefits credited twice monthly instead of once monthly, in hopes that food will more easily reach to the end of the month. With careful attention to research ethics, one may hope that pursuing stronger random-assignment research designs would best serve the original admirable purpose of using food insecurity and hunger measurement to assess and improve antihunger programs.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Manuscript received 28 September 2006. Initial review completed 13 October 2006. Revision accepted 21 November 2006.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTIntroductionLITERATURE CITED |
|---|
1. Eisinger P. Toward an end to hunger in America. Washington: Brookings; 1998.
2. National Research Council, Committee on National Statistics, Division of Behavioral and Social Sciences and Education. Food insecurity and hunger in the United States: an assessment of the measure. Washington: National Academies Press; 2006.
3. Nord M, Andrews M. Reducing food insecurity in the United States: assessing progress toward a national objective. Washington: USDA, Economic Research Service; 2002. Food Assistance and Nutrition Research Report No.: 26–2.
4. Wilde PE. The uses and purposes of the USDA food security and hunger measure. In: Workshop on the measurement of food insecurity and hunger. Washington: National Research Council; 2004.
5. Herman DR, Harrison GG, Afifi AA, Jenks E. The effect of the WIC program on food security status of pregnant first-time participants. Fam Econ Nutr Rev. 2004;16:21–30.
6. Nord M, Romig K. Hunger in the summer: seasonal food insecurity and the National School Lunch and Summer Food Service programs. J Child Poverty. 2006;12:141–58.
7. The Food Stamp Program [homepage on the Internet]. Alexandria: USDA, FNS; 2006 [cited 2006 Sep 27]. Available from: http://www.fns.usda.gov/fsp.
8. Fox MK, Hamilton WL, Lin BH, editors. Effect of food assistance and nutrition programs on nutrition and health: volume 3, literature review. Washington: USDA, Economic Research Service; 2004. Food Assistance and Nutrition Research Report No.: 19–3.
9. Nord M, Andrews M, Carlson S. Household food security in the United States, 2004. Washington: USDA, Economic Research Service; 2005. Report No.: ERR-11.
10. USDA. FNS. Food and Nutrition Service strategic plan 2000 to 2005. Alexandria (VA): USDA, FNS; 2000.
11. Program Assessment. Food Stamp Program [homepage on the Internet]. Washington: Office of Management and Budget; 2006 [cited 2006 Sep 27]. Available from: http://www.whitehouse.gov/omb/expectmore/summary.10001007.2005.html.
12. Wilde PE, Ranney C. The monthly food stamp cycle: shopping frequency and food intake decisions in an endogenous switching regression framework. Am J Agr Econ. 2000;82:200–13.
13. Shapiro JM. Is there a daily discount rate? Evidence from the food stamp nutrition cycle. J Pub Econ. 2005;89:303–25.
14. Barrett A, Poikolainen A. Food Stamp Program participation rates: 2004. Alexandria (VA): USDA, FNS; 2006.
15. Holben DH. Position of the American Dietetic Association: food insecurity and hunger in the United States. J Am Diet Assoc. 2006;106:446–58.[Medline]
16. Burstein NR, Hamilton WL, Fox MK, Price C, Battaglia MP. Assessing the food security and diet quality impacts of FNS program participation: final report. Alexandria (VA): USDA, FNS; 2005.
17. Gundersen C, Oliveira V. The Food Stamp Program and food insufficiency. Am J Agr Econ. 2001;83:875–87.
18. Gibson-Davis C, Foster M. A cautionary tale: using propensity scores to estimate the effect of food stamps on food insecurity. Soc Serv Rev. 2006;80:93–126.
19. Jensen HH. Food insecurity and the Food Stamp Program. Am J Agr Econ. 2002;84:1215–28.
20. Huffman S, Jensen HH. Do food assistance programs improve household food security? Recent evidence from the United States (revised). Ames (IA): Iowa State University, Center for Agricultural and Rural Development; 2006. Report No.: 03-WP 335.
21. Hofferth SL. Persistence and change in the food security of families with children, 1997–99. Washington: USDA, Economic Research Service; 2004.
22. Ribar DC, Hamrick KS. Dynamics of poverty and food sufficiency. Washington: USDA, Economic Research Service; 2003. Food Assistance and Nutrition Research Report No.: 36.
23. Wilde PE, Nord M. The effect of food stamps on food security: a panel data approach. Rev Agr Econ. 2005;27:425–32.
24. Rose D, Gundersen C, Oliveira V. Socio-economic determinants of food insecurity in the United States: evidence from the SIPP and CSFII datasets. Washington: USDA, Economic Research Service; 1998. Report No.: TB 1869.
25. Daponte BO, Stephens MJ. The relationship between food assistance, the value of food acquired, and household food security. Chicago: University of Chicago, Harris School Working Papers; 2004.
26. Nord M, Newman C. Food stamps and food security: does the size of the benefit matter? Fall Research Conference, Association for Public Policy Analysis and Management; 2004 Oct 28–30; Atlanta (GA).
27. Kabbani NS, Kmeid MY. The role of food assistance in helping food insecure households escape hunger. Rev Agr Econ. 2005;27:439–45.
28. Wilde PE. The food stamp benefit formula: implications for empirical research on food demand. J Agr Resour Econ. 2001;26:75–90.
29. Bartfeld J, Dunifon R. State-level predictors of food insecurity among households with children. J Policy Anal Manage. 2006;25:921–42.
30. Borjas GJ. Food insecurity and public assistance. J Pub Econ. 2004;88:1421–43.
31. Nord M, Kantor LS. Seasonal variation in food insecurity is associated with heating and cooling costs among low-income elderly Americans. J Nutr. 2006;136:2939–44.
32. Bhattacharya J, DeLeire T, Haider S, Currie J. Heat or eat? Cold weather shocks and nutrition in poor American families. Am J Public Health. 2002;93:1149–54.
33. Townsend MS, Peerson J, Love B, Achterberg C, Murphy SP. Food insecurity is positively related to overweight in women. J Nutr. 2001;131:1738–45.
34. Wilde PE, Peterman JN. Individual weight change is associated with household food security status. J Nutr. 2006;136:1395–400.
35. Sarlio-Lahteenkorva S, Lahelma E. Food insecurity is associated with past and present economic disadvantage and body mass index. J Nutr. 2001;131:2880–4.
36. Gibson D. Food Stamp Program participation is positively related to obesity in low-income women. J Nutr. 2003;133:2225–31.
37. Jones SJ, Frongillo EA. The modifying effects of Food Stamp Program participation on the relation between food insecurity and weight change in women. J Nutr. 2006;136:1091–4.
38. Chen Z, Yen ST, Eastwood DB. Effects of food stamp participation on body weight and obesity. Am J Agr Econ. 2005;87:1167–73.
39. USDA. Strategic Plan for FY 2005–2010. Washington: USDA; 2006.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
