THE INSTITUTIONAL DYNAMICS OF POLICY ADMINISTRATION - BUREAUCRATIC INFLUENCE OVER SECURITIES-REGULATION

Theory: A dynamic systems model of administrative politics that (1) re laxes restrictive causal assumptions among the president, Congress, an d an administrative agency, (2) emphasizes aggregate institutional beh avior, and (3) analyzes both anticipated and unanticipated policy beha vior is set forth as a more general means to investigate agency-politi cal relationships. Hypotheses: The main set of hypotheses focus on the aggregate institutional relationships among the president, Congress, and administrative agency. These four possible hypotheses include bure aucratic influence, bureaucratic autonomy, mutual influence, and polit ical influence. The purpose here is to assess the relative merits of c ompeting explanations for bureaucratic behavior: Securities regulation is a per feet area of inquiry to investigate these competing perspect ives since the findings of existing research on this topic are dispara te. Methods: Vector autoregressive time series statistical methods are used to empirically test the dynamic systems model of administrative politics by examining three empirical models of SEC enforcement behavi or in relation to presidential and congressional budgetary preference signals for the 1949-92 annual period. Results: The SEC sends potent s ignals of their own (via their changes in regulatory outputs) that wil l affect the budgetary preferences of its political principles. Howeve r, the SEC's administrative outputs are shown to be somewhat less sens itive to political budgetary preference signals than what one would ex pect. Presidential and congressional budgetary preference signals (wit h respect to the SEC) are interdependent to a sizeable extent, thus ra ising doubts about the separate, independent effect democratic institu tions have over the bureaucracy. The implications of this study sugges t that conventional theories of agency-political relations serve as an oversimplified way to study administrative behavior. Future research on this topic