International courts (ICs) clearly fit the paradigm of delegation examined in this volume. States operating as a collective principal create ICs through a revocable delegation contract; appoint IC judges; and can write or rewrite the mandate and laws that ICs interpret. Principal-agent (PA) theory expects courts to be among the more independent “agents,” intentionally so. As Giandomenico Majone argues, in delegation to enhance the credibility of a principal the “Fiduciary Agent” is made independent because “an Agent bound to follow the directions of the delegating politician could not possibly enhance the commitment” (Majone 2001: 110). Thus intentionally principals allow judges to be fired only for egregious acts unbecoming to their office, and judicial salaries are protected. Still, PA theorists expect states to have substantial tools of control because international judicial terms are short (4–8 years), because international judges may worry about their professional futures including whether or not their term is renewed, and because states can sanction ICs through rewriting their mandate, legislating to reverse their rulings, or through non-compliance.

While these expectations are shared by most PA theorists, studies employing PA theory to analyze ICs have offered contradictory predictions about whether and when we should expect IC autonomy. Geoffrey Garrett and Barry Weingast have argued that the European Court of Justice (ECJ) has far less autonomy than national courts because the ECJ fears re-contracting. They assert that ECJ decisions mainly select among the range of outcomes the most powerful states implicitly want (Garrett and Weingast 1993: 201).