Strategic decision support for project portfolio management in the public sector.

摘要

This study addresses the problem of managing multiple large-scale public sector projects, in particular those whose execution will stretch over several time periods. Public projects are usually financed with funds from several sources, each with restrictions as to how the funds may be used. The funds may be restricted to use on a specific project, to specific types of cost or to be used with matching funds from another source. Also of interest is the option in certain cases to leverage the provided funds by issuing bonds. Additionally, the exact amount of income in future years is often unknown as politics, decisions by other agencies, or the economy can have an effect on the future budget. The objective of public planners is to complete the projects in as timely a manner as possible, however, it is difficult to determine which project should have higher priority when tradeoffs are necessary due to budget restrictions.; There are several aspects of this problem that make it interesting and difficult. Both integrality requirements and stochastic elements are present, making modeling the problem and obtaining a solution more difficult. The integer nature of the problem is due to the requirement to choose exactly one schedule for each project under consideration. The stochastic elements come from the unknown nature of the future budgets. The methodologies presented here are designed to analyze situations in which multiple projects are being simultaneously managed over multiple time periods. They are related in structure to three classic operations research problems, the capital rationing problem, the resource constrained project scheduling problem, and the stochastic financial portfolio problem. Several models, each addressing different aspects of the problem, are presented here. Integer programming, column generation and multistage stochastic programming with recourse are the techniques used in the development of this set of decision support models.; This study was motivated by a large public sector agency application that involved multiple related construction projects with severe funding constraints and complicated funding source restrictions. Data from this case problem is used to illustrate the application of these methods.