This was the first article explicitly on the theory of agency published in a regular, i.e., nonproceedings, issue of a journal in social science. The paper presents a fiduciary function model of policing in agency, with an application to attempts to influence regulatory performance by policing the behavior of regulators. Four types of agents - the pure fiduciary, lexical fiduciary, lexical self-interest agent, and pure self-interest agents - are identified. The paper notes that the rational principal would not police his agent if he did not expect a net gain from the attempt; this is one of the key logics of agency theory. The paper notes the effects of the fiduciary norm in economizing on specification and policing (agency) costs. An apparent paradox can occur when policing the agent appears to lower rather than increase the return to the principal. In other words, agent fidelity does not necessarily correlate with the level of principal return. In the context of public regulation, this can take the form of producing a more honest or better-behaved regulatory agent in a government that produces a poorer return to the public interest. The posted version of this article contains some corrections of errors/omissions introduced by the journal's publisher in the publication process.