Food Stamp Use Among the Elderly: Evidence From Panel Data

Year: 2008

Research Center: The Harris School of Public Policy Studies, University of Chicago

Investigator: Levy, Helen

Institution: University of Michigan

Project Contact:
Helen Levy
Institute for Social Research
University of Michigan
426 Thompson Street
Ann Arbor MI 48106-1248
Phone: 734-936-4506


About half of all households that are eligible for food stamps but do not use them include an elderly member. Takeup of food stamps is much lower among the elderly than among younger adults and children; only about 30 percent of the elderly who are eligible take up benefits compared with an overall takeup rate of 59 percent among all eligible individuals. Previous research has found that failure to take up benefits by the elderly cannot be explained by behavioral factors, such as differences in expected benefits or obstacles to enrollment (for example, functional limitations). Thus, low takeup among the elderly, both in absolute terms and relative to younger individuals, remains a puzzle.

Understanding food stamp dynamics among elderly individuals is a high priority for at least three reasons. First, many of the elderly will be at risk of needing benefits at some point. Although the poverty rate among the elderly at any point in time is now only about 10 percent, 40 percent of individuals will experience a spell of poverty between the ages of 60 and 90. This fraction is considerably higher for subgroups, including Blacks and individuals with low education levels. Thus, the program has the potential to benefit a significant number of elderly individuals—if they participate. Second, low takeup may be a warning flag that even those who receive benefits face obstacles to signing up, and the program could be more efficient if these obstacles were eliminated. Third, as the Baby Boomers age and the size of the elderly population increases significantly, accurate projections of food stamp spending require that we understand whether the low takeup rates among the current elderly are likely to continue. If relatively low takeup among today’s elderly is due to differences across cohorts rather than to age, food stamp expenditures for the elderly may increase dramatically once the Baby Boomers, rather than the children of the Depression, are deciding whether or not to sign up for the program.

This study uses panel data from the 1992 through 2006 waves of the Health and Retirement Study (HRS) to analyze patterns of food stamp use and takeup (that is, use conditional on eligibility) over time and across cohorts, as well as the longitudinal predictors of food stamp use and takeup in multivariate models. The HRS is a nationally representative longitudinal study that collects information every 2 years on income, labor supply, program use, health, and other characteristics of individuals over the age of 50. This study uses HRS data on approximately 20,000 individuals born between 1923 and 1953 who are observed for as many as eight survey waves. The data are used to construct age-specific rates of food stamp use and takeup for four birth cohorts: the Children of the Depression (born 1923 to 1930), the original HRS cohort (born 1931 through 1941), the War Babies (born 1942 through 1947), and the Early Baby Boomers (born 1948 through 1953). About 13,000 of these individuals are observed for at least three consecutive waves, and data on this subset of individuals are used to analyze transitions over time between program use and program nonuse, including periods when individuals are eligible for benefits but not receiving them. Finally, multivariate models are used to estimate the determinants of food stamp use, poverty, and food stamp takeup. These models include as covariates age, race, education, marital status, health status (presence of serious health conditions and functional ability as measured by Activities of Daily Living and Instrumental Activities of Daily Living), work status, and household composition. Results are compared for models with and without an individual “fixed effect” that controls for the presence of unobservable time-invariant characteristics.

The analysis finds no real differences across birth cohorts in the use or takeup of food stamps in old age for those born between 1923 and 1953. In particular, the Early Baby Boomers (born between 1948 and 1953) are no more likely to use food stamps in their fifties than were earlier cohorts. Thus, any increase in program use by the elderly as a result of higher takeup among more recent cohorts is at least a few decades off, rather than being just around the corner.

The analysis of transitions over time finds that food stamp use among the elderly and near-elderly is quite persistent in the sense that nearly all of those who use food stamps at all during a calendar year receive them in every month of that year. However, over a longer time horizon, there is quite a bit of movement on and off the program in this population. Most of those who use food stamps at some point during a 6-year interval will not use them all 6 years. Transitions on and off the Food Stamp Program typically involve a period of eligibility without receiving benefits. Most spells of eligible nontakeup do not end in program takeup, but most spells of program use begin or end with a period in which available benefits are not taken up. In this sense, periods of eligible nontakeup are a stepping stone on and off the Food Stamp Program.

Finally, multivariate models of food stamp use and takeup yield quite different results, depending on whether or not they include an individual fixed effect. In particular, while regressions without a fixed effect (FE) suggest that older age is associated with lower takeup, individual FE models suggest that the elderly are more likely to take up benefits as they age. Work, marriage, and good health, which are strongly associated with lower program takeup in ordinary least squares models, do not significantly predict food stamp takeup in FE models. The fact that individuals who retire or lose a spouse—both events that increase the risk of poverty among the elderly—do not take up food stamp benefits suggests a possible intervention to protect the elderly from economic hardship. Both retirement and the death of a spouse are events that are, in theory, known to the Social Security Administration because they affect the receipt of Social Security benefits. If these events triggered automatic receipt of information about food stamps, more eligible beneficiaries might take advantage of benefits available to them. Understanding the potential of food stamps to mitigate the hardships potentially facing retirees, widows, and widowers is a high priority for future research.

Direct inquiries about this study to the Project Contact listed above.