consortiums

PROJECT 3

DISTRIBUTION AND ASSIGNMENT

Laura Razzolini,
School of Business Administration,
University of Mississippi

Proposed Grant $68,494

Abstract:
Large and complex organizations face a continuous challenge of allocating common shared resources efficiently and effectively. The Navy, one of the largest and complex organizations, also faces the continuous challenge to allocate its resources in the most efficient and effective manner possible. In research previously funded by NSF and ONR, the P.I. has investigated

the use of a particular mechanism, the “serial cost sharing rule,” to allocate scarce resources in a way that accurately reflects the underlying needs. Whenever the supply of the shared good is not limited (i.e., the total demand can be satisfied), the serial cost sharing rule has been suggested as a method to use to discover the individuals’ true demands.

Whenever the supply of the shared good is limited (i.e., the total demand cannot be satisfied), different mechanisms need to be considered since now two issues arise: 1) how to allocate the costs among different individuals, and 2) how to select some users among a too large set of demanders. The first mechanism we propose to consider and analyze is a randomized version of the serial cost sharing rule: the Probabilistic Serial mechanism (PS). Theoretically, this mechanism needs to be extended in the more general case, in which agents have a demand for the service which may vary over time, as other agents belonging to his/her same unit or agency are admitted to training.

Alternatively, we may want to be able to discriminate among recipients of resources. In this case, a Priority Rationing mechanism (PR) is needed. Mechanisms used in real life as FiFO (First-in-First-Out) or FiLO (First-in-Last-Out) are examples of PR mechanisms.
Each one of these probabilistic and priority mechanisms need to be fully formalized, analyzed and compared. The incentive property of these mechanisms needs to be verified. In this project we propose to perform the following tasks.

· The Probabilistic Serial mechanism has been suggested by Moulin as a possible rule to ration and allocate costs among users. This mechanism needs to be characterized for a more general case where agents have demands for the service that vary over time. The incentive property of this mechanism will be verified.
· The PS mechanism will be evaluated experimentally. We will need to verify if it is implementable, if the strategic properties hold in practice, and evaluate which version of the extended serial cost sharing rule is better and easier to implement in the field.
Other rationing mechanisms are available in the literature besides the PS mechanism; e.g., the proportional rationing mechanism, priority rationing mechanisms such as the fair queuing mechanism, or queuing protocols as FiFO or FiLO. None of these mechanisms have been validated experimentally. We propose to perform experiments to test whether each mechanism is implementable and whether the normative properties hold in practice. Comparisons in terms of welfare and fairness in performance are also needed and will be performed. We will provide answers to the following open questions: Which mechanism minimizes the failure rate? Which mechanism maximizes collective surplus?