DECENTRALIZEDDECISIONMAKING IN DYNAMIC TECHNOLOGICAL SYSTEMS The Principal-Agent Paradigm

Operations Management (OM) is concerned with the study of technological systems that produce good or services through the repetitive execution of standardized tasks. Although most of these systems are decentralized with an owner (or principal) delegating operational control to one or more noncooperative decision makers (or agents), traditional OM models suppress that feature for analytical tractability. Yet, the behavior of the noncooperative agents is also important and should be captured in OM models. Numerous recent research efforts have attempted to do that, but to maintain tractability they suppress most of the system complexity which is the landmark of OM models; see Anupindi, Bassok, and Zemel (1999), Cachon and Lariviere (1997), Lippman and McCardle (1997), Van Mieghem (1998). In addition, the few notable exceptions that capture both the processing dynamics and the agents’ behavior utilize sophisticated models whose tractability relies on specialized assumptions; see Bassok and Anupindi (1999), George (1999), Hall and Porteus (1999), Lee and Whang (1996). Decentralized systems are plagued by the uncoordinated actions of the multiple noncooperative agents and, thus, can be inefficient. However, the agents’ actions can be coordinated either by adopting speciÞc incentive contracts that motivate them to behave in a way that is desirable from the system’s perspective, or else by promoting competition between them. In either case, to effectively manage such systems, one