Legislative control, bureaucratic characteristics, and discretion: The shaping of policy outcomes in state welfare bureaucracies

As the scope and complexity of the United States government grew during the twentieth century, elected officials responded by increasingly delegating the refinement and application of policy to administrative entities. The exercise of discretion by public administrators that is necessarily part of the resulting implementation process is subject to legislative, judicial, and executive constraints. Nonetheless, discretionary decision making in government bureaucracies further shapes elected officials' policy intentions. The present study investigates outcomes resulting from bureaucratic discretion using the case of Aid to Families with Dependent Children (AFDC). During the fourteen years prior to its end in 1996, political leaders increasingly favored work requirements and tighter program rules as ways to move able-bodied adults off the program rolls and into the workforce, and state welfare bureaucracies were called upon to implement these policy goals. This research focuses on state-level outcome measures that reflect levels of applicant access, participant termination, and work requirement enforcement. Models based on two bodies of theory are developed to explain these outcomes. The legislative control model holds that discretionary decision making is shaped by the beliefs and actions of elected officials. The bureaucratic factors model suggests that characteristics of agencies and their environments also systematically influence decisions and outcomes. These two models are tested empirically to investigate the relation between these forces and outcomes in bureaucracies with high levels of worker-citizen contact. Pooled time series analysis is used to analyze data from the fifty states from 1982 to 1995. Findings provide mixed support for the legislative control model. More liberality in state government is positively related to both higher application approval rates and lower termination rates. Stringent reform measures appear to exert their intended influence only on that outcome to which they were most closely related, moving recipients out of the program. The findings provide little support for the bureaucratic factors model. Unexpected significant findings are also discussed.