A Theory of Rule Development

This paper develops a model with endogenously coarse rules. A principal hires an agent to take an action. The principal knows the optimal state-contingent action, but cannot communicate it perfectly due to communication constraints. The principal can use previously realized states as examples to define rules of varying breadth. We analyze how rules are chosen under several assumptions about how rules can be amended. We explore the inefficiencies that arise and how they depend on the ability to refine rules, the principal’s time horizon and patience, and other factors. Our model exhibits path dependence in that the efficacy of rule development depends on the sequence of realizations of the state. We interpret this as providing a foundation for persistent performance differences between similar organizations and explore the role of different delegation structures in ameliorating the effects of bounded communication. ∗Ellison: Massachusetts Institute of Technology, Department of Economics, E52-380a, 50 Memorial Drive, Cambridge MA 02142. E-mail: [email protected]. Holden: Massachusetts Institute of Technology, Sloan School of Management, E52-410, 50 Memorial Drive, Cambridge MA 02142. E-mail: [email protected]. This work was supported by MIT Sloan’s Program on Innovation in Markets and Organizations, NSF grant SES-0550897, and the Toulouse Network for Information Technology. We are deeply indebted to Bob Gibbons for numerous discussions and suggestions. We also thank Oliver Hart, Bengt Holmstrom, Andrei Shleifer, Andy Skrzypacz, Oliver Williamson, and seminar participants at Chicago GSB, Harvard, MIT, UC Berkeley, Columbia, and UCLA for helpful comments.