Murat Bayiz

We study the problem of the manager of a project consisting of multiple sub-projects

or tasks which are outsourced to different subcontractors. The project manager earns

more revenue if the project is completed faster, but cannot observe subcontractors’ effort,

only the stochastic duration of their tasks. We outline how to determine the optimal

linear contracts for general networks with normally distributed activity durations using

an approximation from Clark (1961). We then derive the optimal linear contracts for the

case with n multiple tasks in series or two tasks in parallel. We discuss when incentive

contracts lead to bigger performance improvements compared to the fixed-price contracts often encountered in practice. We characterize how the incentive contracts vary with the subcontractors’ risk aversion and cost of effort, the marginal effect of subcontractor effort, and the variability of task durations. This dependence is sometimes counter-intuitive. For

parallel tasks, if the first agent’s task is on the critical path and his variability increases,

the project manager should induce the first agent to work less hard and the second agent to

work harder. Our numerical analysis of more complex networks suggests that the structure

of project networks affects the optimal contracts and that the value of incentive contracts and the value of information are higher in projects with a dominant critical path than in

projects with many parallel critical paths.