Hierarchy, Learning and Dynamic Moral Hazard

. Hierarchy in organizations has long been regarded as a source of moral hazard problems and an indication for some inherent difference between workers in different hierarchical positions. However, hierarchies may actually be a response to moral hazard problems. If past effort, rather than time or outcomes determine the correct expectations on current efficiency, single agent mechanisms distort incentives and reduce payoffs and efficiency. Organizations that limit themselves to single agent structures may be forced to allocate all the dynamic gains to the agent, reducing efficiency and profits. Hierarchies limit the extent to which a single agent can “game” the system by automatically transferring tasks that require stronger incentives than expected to another agent. This enables organizations to learn from the failures of lower levels in the hierarchy and improve incentives for higher levels. This paper solves a simple but general model of past effort effects on current productivity and develops an incentive based hierarchy. The characteristics of the hierarchy and the effects of knowledge transfer costs and production primitives on the organizational structure and players profits are identified. By turning a dynamic problem into a sequence of static problems, hierarchy is an extremely simple yet effective way for organizations to contain dynamic moral hazard issues. The optimal hierarchy is described by a relatively small set of scalars - stopping conditions for each layer of the hierarchy. While simple, I find that for a large class of problems and possible multi-agent mechanisms, hierarchy is the most profitable solution for the organization.