Food assistance benefits, such as food stamps and the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), exist, in part, to mitigate the year-to-year variability in worker earnings and in household incomes. Several previous studies have found that households with more volatile incomes are more likely to be food insufficient. The amount of year-to-year variability in U.S. household incomes is substantial, although there is debate as to whether this variability has been increasing in recent decades. Increases in imputation rates across Survey of Income and Program Participation (SIPP) panels can create a mechanical, and incorrect, increase in measured income volatility over time. This study explores the relationship between household income volatility and food insecurity, the trend in this relationship, and whether these measured relationships are affected by the rising rates of income imputation in the SIPP.
This study uses data from the 1991, 1993, 1996, 2001, and 2004 SIPP panels. Income volatility is measured by using an indicator of total (pre-tax) household income, which is the sum of earned and nonearned income for each household member, declining by 25 percent or more. Our main outcome variable of interest is food insufficiency, which is measured in the SIPP based on respondents’ answers to the following question: “Getting enough food can also be a problem for some people. Which of these statements best describes the food eaten in your household in the last four months?” Households are coded as being food insufficient if they respond that there is “sometimes not enough to eat” or “often not enough to eat.”
The study determines the relationship between income volatility and the probability that a household reports being food insecure, controlling for its level of income and other household characteristics. It then determines whether this relationship changes when we drop observations with imputed income. It measures these relationships among all households, among households below 130 percent of the Federal poverty line (FPL), and among households below 200 percent of FPL. It also determines the trend in the relationship between income volatility and food insufficiency and whether this trend changes when households with imputed incomes are excluded from the analysis.
Rates of income imputation in the SIPP are large and growing. In the 2004 panel, for example, roughly 60 percent of observations have some source of household income imputed over a 2-year period. This study shows that the inclusion of imputed observations leads to a serious understatement of the association between income drops and food insufficiency, both among
households of all income levels and among low-income households. It also shows that this association is getting smaller over time and that including imputed observations partially masks this decline (reducing its apparent rate of decline by half).
In particular, the association between income volatility (income drops) and food insufficiency is about 75 percent larger than one would believe using all of the observations provided by the Census Bureau, including the imputed observations. This association has been declining over time—a decline that is twice as large as one would believe if one used the imputed data. We strongly advise caution when using imputed observations and examining changes in income in the SIPP.
Direct inquiries about this study to the Project Contact listed above.