Regulatory agencies have many supervisors. Congress gives them life, appropriates their money, and conducts oversight of their activities. The president appoints and can remove their senior officers, occasionally instructs them, and conducts review of many of their important actions before they take effect. Federal courts, too, supervise agencies. Statutes authorize federal courts to entertain complaints about agency action and, in the course of hearing those challenges, federal courts evaluate the legal validity of agency action. If a regulatory agency goes astray, each of these institutions has some ability to nudge it back to the proper path.

These supervisors have distinct tools at their disposal to supervise agencies. Congress, on paper, has nearly limitless authority over agencies, constrained only by the Constitution. Assuming the approval of the president, or a veto-proof majority, congressional power to legislate and appropriate can be deployed to control an agency in any number of ways. It can specifically tell an agency what to do, it can generally instruct an agency, it can prohibit future action, or it can reverse past action (or all four at once). The president, by contrast, has much more limited authority to initiate agency action or to alter the agenda of an agency, but there are many crucial “implementation” decisions that agencies make that are not clearly dictated by statute, and in that space, the president can seek to work his will through his appointments and his supervision. Compared with the institutions set forth in the first two articles of the Constitution, the federal courts identified in Article III seem to have the most limited authority to supervise agency action. They are purely reactive, permitted to intervene only when a challenger to agency action seeks to invoke the authority of the federal courts. Federal judges review a tiny slice of agency actions and they are not the authors of the playbook that guides their review – the judges are guided by statutes.