The Role of the Social Safety Net in Mitigating Economic Inequality in California
*Names in bold indicate Presenter
This research has to-date been directed solely at poverty measurement. We suggest that the methodology also holds promise for enriching our assessments of economic inequality. In this paper, we develop metrics of inequality in resources, using the SPM resource concept that reflects after-tax cash income and near-cash supports like food and housing assistance. Our measure of resource inequality has a number of advantages over measures based on pre-tax cash income (Piketty and Saez, 2004). The estimates are more comprehensive and accurate in the sense that all major safety net programs are included. Pre-tax cash income includes TANF and SSI benefits but not EITC benefits, for example. Because our SPM-style measure for California was developed using the American Community Survey, we also have the advantage of a large sample with which we can investigate inequality across more fine demographic subgroups than is possible with studies based on the smaller Survey of Income and Program Participation or Current Population Survey (Congressional Budget Office, 2011; Moffitt, 2013). Our ACS-based methodology also uses restricted administrative data on SNAP and TANF participation to correct for survey underreporting which is substantial and varies by race (Bohn, Danielson, and McConville, 2014, Meyer, Mok and Sullivan, 2009), providing increased accuracy.
We address three main research questions in this paper. First, how much do resource inequality measures differ from traditional income inequality ones, especially comparisons made between the middle and the bottom of the distribution? Second, by how much does each safety net program (and all together) reduce inequality? Third, we explore how the distribution of resources would shift given changes in government programs. We create a set of reasonable changes to safety net policy based on recent state and federal policy discussions (e.g. EITC expansion, SNAP participation increase and decrease in churn). Where possible, we include behavioral responses to policy changes based on the literature (e.g., Schanzenbach and Hoynes, 2012; Eissa and Liebman, 1996). We will estimate upper and lower bounds by varying behavioral assumptions in the model (e.g., work responses and take-up of expanded programs).