Incomplete Incentive Contracts under Ambiguity and Complexity

Abstract

Many novel projects are characterized by ambiguity (impossibility to recognize all influence variables and to foresee all possible events) and complexity (interaction of many performance influence variables, making the overall performance difficult to estimate). Two fundamental approaches to project management under these conditions have been identified: Selectionism, or pursuing multiple approaches independently of one another and picking the best one ex post, and trial & error learning, or flexibly adjusting to new information about the environment as it emerges. While the actions to be taken under the selectionist approach can be defined at the outset, and thus standard contracting theory applies, trial & error learning involves taking actions after ambiguity has been resolved, making it inherently difficult to set incentives for managers. Actions and targets cannot be specified at the outset, since they would no longer be optimal at the time the actions should be executed. In a search model in a complex performance landscape, this paper first shows that trial & error learning is more attractive than selectionism when ambiguity and high complexity combine. Second, for this situation of trial & error learning, we construct incomplete contracts between a principal (e.g., the firm) and an agent (e.g., a project manager) that can re-instate optimal incentives for the agent. This is achieved by a priori defining time points and aspects of re-negotiation, depending on what each party learns. As the project manager, as an employee, is ambiguity averse, he must be protected from unforeseeable variations in his compensation. The principal, in contrast, is willing to accept ambiguity, and the incomplete contract offers him a means to optimally re-direct the agent's actions in return or insuring the agent against payment ambiguity.