The extent to which safety net programs mitigate or amplify the income instability caused by changes in employment and family composition, while not well understood, is fundamental to future federal, state, and local policy making and program administration. Ideally, families would exit safety net programs after a permanent increase in income, but the reality of low-income family life is that changes in employment or family structure are rarely permanent. Even consistently employed less-educated workers face substantial instability in work hours. The dynamic nature of low-income family life is not captured in most studies of the effectiveness of safety net programs, which generally examine point-in-time measures of employment, income, or caseload size.

This study uses state policy data merged with two panels of the Survey of Income and Program Participation (SIPP) to examine how state-level choices about five key safety net programs—TANF, SNAP, Medicaid/SCHIP, Unemployment Insurance (UI), and child care subsidies (CCDF)—affect the economic stability of households. The policy database was compiled from reliable, publicly-available sources including The Urban Institute’s Welfare Rules Databooks, reports from the Kaiser Commission on Medicaid and the Uninsured, yearly USDA Food and Nutrition Service State Options Reports, and Department of Labor compilations of state UI rules. The analysis focuses on policies that are plausibly related to the stability of income, either directly (through income supports) or indirectly through the effects on employment and earnings.  These policy rules are generally related to eligibility determination processes, recertification processes, restrictions of the length of receipt, and earned income disregards. The dependent variables in this study capture meaningful variability in household economic circumstances over time. Intra-year income variability is measured using arc percent change (APC). Unlike a basic percent change measure, APC is symmetric (changes of the same size in opposite directions produce the same arc percent change) and defined even if income is zero in one of the time points.  We also measure the frequency of large income changes by counting the number of changes greater than 33% of average monthly income. We examine the associations between state policy choices and total income (in)stability, as well as disaggregating the effects on the main components of income, earned income and means-tested income.