Coordination and incentive contracts in stochastic project management

In this dissertation, I consider coordination and incentive contracting problems in projects in which activity durations are random variables. In the first chapter, I examine how resource allocations are affected by the variability in activity times. I study the case where the project, consisting parallel subprojects, has a due date and the project manager is rewarded for early completions or penalized for delays. I assume that the project manager has full information and allocates resources to two subcontractors whose activity durations stochastically decrease with the level of resources allocated to them. Results show that the project manager does not necessarily allocate more resources to the subcontractors with higher activity duration variability.; In the second chapter, I study the incentive contracting problem for a project consisting of two sub-projects or tasks that are outsourced to different subcontractors, now explicitly introducing information asymmetry. The project manager earns more revenue from the project if it is completed faster, but she cannot observe the subcontractors' effort levels, only the stochastic duration of their tasks. I derive the optimal linear incentive contracts and compare them to the fixed-price contracts often encountered in practice. I 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. I find that this dependence is sometimes counter-intuitive; the effect of a decrease in the first activity's completion time variability on the payment to the first agent depends on whether that activity is likely to be on the critical path.; The third chapter extends the incentive contracting problem from the second chapter to a general project network. For the deterministic environment, I show that the incentive contracting problem with a linear cost function can be trans formed to a time-cost trade-off problem and that with a convex cost function can be transformed to a minimum cost network flow problem. In the stochastic environment, I take the activity durations to be exponentially distributed random variables, which enables us to transform the project network to a continuous time Markov chain with an absorbing state. I show how to compute the optimal linear contract for an arbitrary network, and derive closed form solutions for contract parameters for a simple project network with two parallel activities followed by one serial activity.