סמינר באסטרטגיה וכלכלת עסקים

Daniel Fershtman, Department of Economics, Northwestern University. 

I study dynamic delegation of heterogeneous projects to agents with diverse capabilities. 

Each agent’s (e.g., division managers, employees) true ability to carry out projects varies over time based on his expertise and private idiosyncratic compatibility with the specifics of the current project. 

The principal’s (e.g., headquarters, management) ability to credibly provide incentives in order to delegate efficiently hinges on the degree of specialization across agents. 

Efficiency - where each project is assigned to the agent best suited for it - is attainable if and only if specialization does not exceed a threshold. 

If specialization is sufficiently high, communication breaks-down entirely. 

The derivation of a necessary and sufficient condition for efficiency, at fixed discounting, enables constructing a simple class of delegation rules that are efficient whenever any rule is, and deriving the key properties of such rules. 

These properties shed light on the potential benefits or drawbacks of certain management practices in the absence of monetary incentives. 

I establish an equivalence between ex-post equilibria – in which agents’ ex-post incentive constraints are satisfied in each period – and a natural class of equilibria in which delegation is driven by past performance, but does not condition directly on past communication. 

The analysis also studies optimal delegation when the principal is unable to discriminate between the agents, and characterizes the cost associated with this inability.