Competitive Equilibria in Combinatorial Exchanges with Financially Constrained Buyers:Computational Hardness and Algorithmic Solutions

Abstract: Advances in computational optimization allow for the organization of large combinatorial markets. We aim for allocations and competitive equilibrium prices, i.e. outcomes that are in the core. The research is motivated by the design of environmental markets, but similar problems appear in energy and logistics markets or in the allocation of airport time slots. Budget constraints are an important concern in many of these markets. While the allocation problem in combinatorial exchanges is already NP-hard with payoff- maximizing bidders, we find that the allocation and pricing problem becomes even $\Sigma_2^p$-hard if buyers are financially constrained. We introduce mixed integer bilevel linear programs (MIBLP) to compute core prices, and propose pricing functions based on the least core if the core is empty. We also discuss restricted but simpler cases and effective computational techniques for the problem. In numerical experiments we show that in spite of the computational hardness of these problems, we can hope to solve practical problem sizes, in particular if we restrict the size of the coalitions considered in the core computations.