We plan to construct a theoretical bridge between classical dynamic allocation models used in Operations Research/Management Science, and between the modern theory of mechanism design. Our theoretical results will generate insights for the construction of applied pricing schemes and testable implications about the pattern of observed prices. The Economics literature has focused on information and incentive issues in static models, whereas the Operations Research/Management Science literature has looked at dynamic models that were often lacking strategic/ informational aspects. There is an increased recent interest in combining these bodies of knowledge, spurred by studies of yield management, and of decentralized platforms for interaction/ communication among agents. A general mechanism design analysis starts with the characterization of all dynamically implementable allocation policies. Variational arguments can be used then to characterize optimal policies. The research will focus on models with multidimensional incomplete information, such as: 1) Add incomplete information to the dynamic & stochastic knapsack problem; 2) Allow for strategic purchase time in dynamic pricing models; 3)Allow for competing mechanism designers. The ensuing control problems are often not standard and require special tools. An additional attack line will be devoted to models that combine design with learning about the environment. The information revealed by an agent affects then both the value of the current allocation, and the option value of future allocations. We plan to: 1) Derive the properties of learning processes that allow efficient, dynamic implementation; 2) Characterize second-best mechanism in cases where adaptive learning and efficiency are not compatible with each other.
Call for proposal
See other projects for this call