Defining welfare spells. Coping with problems of survey responses and administrative data

The authors explore how to define a welfare spell and how well surveys measure welfare spells. By comparing survey and administrative data from the Work Pays Demonstration Project in California on the receipt of Aid to Families with Dependent Children (AFDC), they find that a substantial amount of administrative churning occurs in administrative data. Through a mixing model of several break lengths, the authors find that a single definition of a break in welfare is not applicable to all respondents. Additionally, it appears that there is substantial variation in the break lengths respondents utilize. Finally, the authors show that the complexity of defining an accurate break in spells creates difficulties for detecting biases in survey responses.