I present new evidence on who uses SNAP, and when, with particular emphasis on the role of the macroeconomy and public policy changes.  Using data from both the Current Population Survey and Food Stamp Quality Control System, I begin by documenting changes in the socioeconomic composition of the caseload across age, gender, race, employment status, and family structure.  Eligibility for the program depends on passing three tests—a gross income test (less than 130% of the federal poverty line), a net income test (gross income less deductions must be less than 100% of poverty line), and an asset test (both vehicles and liquid assets).  The gross income test is waived for disabled persons and those over age 60, and the asset test is also higher for those groups.  With the aging of the population, and rise in disability, it is possible that the program has shifted from an income maintenance program for poor parents and their children to one that serves primarily adults, and increasingly older adults. 

If net income is zero then the recipient qualifies for the maximum benefit, while if net income is positive then the monthly benefit is reduced by 30% for every dollar of net income. Although SNAP was designed as a supplement for food purchases, I quantify changes in the fraction who is receiving the full benefit and thus in effect reliant on SNAP for all food purchases.  In addition, SNAP units are closer to tax units than they are to households in the sense that an unmarried couple with children will get more SNAP because they are two SNAP units, than if they were married and thus became one SNAP unit. With changes in family structure away from marriage and toward cohabitation, SNAP may have become more valuable at the household level and I will document how much this translates into greater food spending power for the household.  

In the second half of the paper I examine whether the level and shifting composition of the SNAP caseload is due to demographic factors (i.e. changes in family structure), macroeconomic factors such as increasing wage inequality and volatile business cycles like the Great Recession, or policy reforms such as the 1990s welfare reform and the expanded outreach of the last decade.  As part of this, tests will be conducted whether participation is now driven more by secular trends such as stagnant real wages and rising wage inequality and less by short-run changes in the business cycle.  If true, then going forward SNAP may ultimately function less as an automatic stabilizer and more as a permanent income (wage) support program.  I will discuss the potential policy implications for SNAP as anti-recession tool versus work support.