Discretion is at the center of most accounts of bureaucracy. Legal scholars in particular have called for agency supervisors, such as Congress, the courts, or the President, to tame that agency discretion. Strangely absent from these accounts is a ubiquitous phenomenon: administrative agencies routinely limit their own discretion when no source of authority requires them to do so.
This Article aims to create a category of such “self-regulation” and argue that scholars have been mistaken to ignore it. It first defines the category of self-regulation, including the feature of administrative law that makes the category interesting, which is that courts will under certain conditions force agencies to follow their self-regulatory measures. That feature of administrative law gives agencies some capacity – albeit limited – to credibly commit to the stability of their self-regulatory measures. The Article next identifies what an agency can accomplish by self-regulating. That includes controlling delegation within the agency, inducing reliance by outside parties, protecting agencies’ policy choices into the future and from political interference today, and producing public goods like information or reputation. Finally, the Article concludes by demonstrating that serious study of self-regulation could change the way we understand agencies and the need for and utility and wisdom of various controls on their behavior.