On Managing Stochastic Decentralized Projects

Thesis (Ph.D.)--University of Washington, 2016-06

Managing decentralized projects effectively is a critical issue today as projects have become increasingly complex, costly, and strategically important (especially IT and product development projects). In this dissertation, we analyze a decentralized project that is composed of n serial stages with stochastic durations; the project is planned, organized, and funded by a client organization that contracts the work at each stage to independent subcontractors. The focus of this dissertation is to build a game theory framework to examine the performances of various types of contracts. In chapter two, we propose an exponential incentive payment contract. Our goal is to maximize the client’s expected discounted profit. Our proposed contract reflects the convex time-cost trade-off that is well known in the project scheduling literature. We show that this type of contract dominates a fixed price contract with respect to expected client’s profit and schedule performance, regardless of payment timing considerations. Using a piece-wise linear approximation, we show that our contract is a generalization of an incentive/disincentive contract that is frequently used in practice. We show how our contract can be used to find the optimal due date and penalties/bonuses in an incentive/disincentive contract. We compare this contract with several variations and discuss implications for both the client and subcontractors. In chapter three, we analyze the case where both the client and subcontractors incur an overhead/indirect cost, an important cost element in projects, in addition to resource related direct costs and possible penalty/delay costs. We propose an Exponential Incentive Contract (EIC) that coordinates a decentralized project with risk neutral subcontractors under discounting. In the case where discounting is minimal and can be neglected, we show that the first order Taylor series approximation of EIC is a Linear Incentive Contract (LIC) and it coordinates the decentralized project with risk neutral and risk-averse subcontractors. We also discuss how EIC can be better implemented in practice by approximating it by contracts that are commonly used. In chapter four, we numerically analyze the cost plus contract and show that the cost plus contract actually creates more uncertainties as it led subcontractors to have non-unique optimal work rates. The performance of the cost plus contractor falls in between that of a fixed price contract and an incentive contract regarding the level of project coordination, client’s and subcontractors’ expected profits. Moreover, in a dynamic setting where subcontractor work rates can be adjusted without incurring significant additional costs, the cost plus contract optimal work rate will converge to a fixed price contract optimal work rate.