This study investigates the relationship between monthly income volatility and food assistance receipt among low-income U.S. households since the early 1990s as well as differences in this relationship by geographic (South versus non-South). Previous research shows a substantial rise in monthly income volatility among the lowest income U.S. households between 1992 and 2003. This rise (which occurred during a period of significant reform of welfare policy) was largely a result of a shift away from relatively stable welfare income to less stable earnings from the labor market. Additionally, researchers have shown that income shocks are an important determinant of food insufficiency and food insecurity. Controlling for average income levels among low-income households (and a rich highly detailed data set of household characteristics), negative income shocks are associated with significant increases in household food insufficiency. Other prior research has found that income volatility is positively related to numerous other non-food related measures of material hardship (for example, trouble paying rent, mortgage or utilities, utility shutoff, eviction, or failure to seek medical or dental care) as well as to self-reported health status of household heads. Thus, households experiencing income fluctuations suffer real and asymmetric shocks to their material well-being along numerous dimensions.
In this context of rising monthly income volatility, this study investigates whether monthly income volatility affects food assistance receipt for low-income households. For example, if program certification and recertification depends on longer term income or is sluggish or costly (for example, in time or effort required), assistance may be relatively unresponsive to short-term income deprivation. Households suffering income volatility or negative income shocks may not be eligible or may not seek eligibility, despite their immediate need. Alternatively, if food assistance receipt depended only on the prior month’s income, this would not be the case. These questions are potentially relevant to eligibility and uptake for a variety of public assistance programs. However, with over 47 million recipients at present and $78 billion in Federal expenditures in 2011, the Supplemental Nutrition Assistance Program (SNAP) takes on special importance as a primary support program for the poor in the United States. Yet, the implications of short-term income instability for food assistance receipt is relatively unstudied and poorly understood to date.
Using nationally representative samples of low-income households from the Survey of Income and Program Participation (SIPP) panels spanning 1991 to 2008, this study investigates the relationship between food assistance receipt and monthly income volatility in greater detail and over a longer period of time than previous authors. In addition, the U.S. Department of Agriculture’s Economic Research Service SNAP Policy database is used to account for the impact of State-level policy changes to the food assistance program over the latter part of this time period. Multivariate logistic regression is used to examine the extent to which income shocks and/or volatility result in changes in food assistance receipt, controlling for income level and a host of relevant controls for household eligibility and need.
A number of findings here are notable. First, positive and negative income shocks have asymmetric effects on immediate food assistance receipt. This is consistent with asymmetry in the food assistance entry and exit processes. Second, with the exception of the 2008 SIPP Panel, income volatility (CV) reduces food assistance receipt for lower-income households below the food assistance gross income eligibility threshold, while the same is not true for their counterparts above the income eligibility threshold. In 2008, income volatility reduces food assistance receipt for households in both income groups. Third, while some State-level policy changes in the post-2001 period are found to have statistically significant influences on food assistance receipt, their inclusion in estimated models has no effect on the income effects estimated here. Fourth, the analysis revealed little evidence of differences in any of the factors studied across Southern and non-Southern households. While the rollout of various food assistance policy changes clearly varied across regions in the post-2001 period, and the effect of policy adoptions was variable across regions in some cases, the overall findings here are comparable Nationwide. Finally, while the relative influence of different income components on food assistance receipt is generally stable, the influence of all income components appears to be diminishing in rather striking fashion across the time period studied. This shift is not attributable to the State-level policy changes accounted for here in the post-2001 period, and it occurs in both Southern and non-Southern States. This finding in particular certainly bears further investigation.