The allocation of a shared resource within an organization

Abstract

Many resources such as supercomputers, legal advisors, and university classrooms are shared by many members of an organization. When the supply of shared resources is limited, conflict usually results between contending demanders. If these conflicts can be adequately resolved, then value is created for the organization. In this paper we use the methodology of applied mechanism design to examine alternative processes for the resolution of such conflicts for a particular class of scheduling problems. We construct a laboratory environment, within which we evaluate the outcomes of various allocation mechanisms. In particular, we are able to measure efficiency, the value attained by the resulting allocations as a percentage of the maximum possible value. Our choice of environment and parameters is guided by a specific application, the allocation of time on NASA's Deep Space Network, but the results also provide insights relevant to other scheduling and allocation applications. We find (1) experienced user committees using decision support algorithms produce reasonably efficient allocations in lower conflict situations but perform badly when there is a high level of conflict between demanders, (2) there is a mechanism, called the Adaptive User Selection Mechanism (AUSM), which charges users for time and yields high efficiencies in high conflict situations but, because of the prices paid, in which the net surplus available to the users is less than that resulting from the inefficient user committee (a reason why users may not appreciate ‘market solutions’ to organization problems) and (3) there is a modification of AUSM in which tokens, or internal money, replaces real money, which results in highly efficient allocations without extracting any of the users' surplus. Although the distribution of surplus is still an issue, the significant increase in efficiency provides users with a strong incentive to replace inefficient user committees with the more efficient AUSM.