Modeling principal-agent feedback in public policy

Of the many debates among students of American politics, no one strikes at the principles of democratic governance more than the controversy over the role of the bureaucracy in public policy decisionmaking, particularly for regulation policies. The primary positive issues revolve around identifying "discretionary" bureaucratic action in the conduct of public policy, estimating the extent to which discretion characterizes the activity of a bureaucratic agency, and determining how the three Constitutionally-specified branches of government (the President, the Congress and the Courts) influence agency behavior. The extant empirical research on the subject has focussed mainly on the influence or control exercised by political principals and/or bureaucratic agents. However, as Hammond and Knott (1992) demonstrate, inferences of institutional control based on empirical findings will be incomplete: empirical models cannot distinguish among substantively different implications of no significant relationship between principal and agent. Further, if both bureaucratic autonomy and political control are a matter of degree, as implied by Hammond and Knott and confirmed in this dissertation, then typologies which attempt to classify political-bureaucratic on using hard-and-fast criteria are of questionable utility. Finally, these models contain specification errors which render them inaccurate representations of the institutionally-grounded public policy decision-making process. In light of this shortcoming, this dissertation serves two purposes. The first is to rectify the specification errors of previous empirical research and construct a more comprehensive empirical model of the policy maintenance process. The second is to redirect empirical policy research toward the more practical aspects ofinstitutional decisionmaking given the problems associated with inferring dominance or control from empirical models. As my findings indicate, the factors that influence both political and bureaucratic decisions are complex enough to resist simple, discrete classification. Further, my findings challenge the positive theoretical framework, the principal-agent model, upon which most formal and empirical models of political-bureaucratic interaction are based.