An Efficient Incentive-Compatible Combinatorial Market Mechanism

Service providers lease bandwidth from owners of individual links to form desired routes. Bandwidth is leased in indivisible amounts, say multiples of 100 Mbps. We study the interaction between buyers and sellers of bandwidth. Within a conventional market, we showed in [8] that a competitive equilibrium exists if agents’ utilities are linear in bandwidth (and money) and they truthfully reveal them, and if the desired routes form a tree. However, strategic agents will not be truthful and the competitive equilibrium will not be realized. To ensure a good outcome among strategic agents, we propose a combinatorial double auction and examine the resulting Nash equilibria. We show that such an equilibrium exists and, more surprisingly, the resulting allocation is efficient. In reality, the players have incomplete information and are risk averse, so we consider the Bayesian-Nash equilibrium under the ex post individual rationality constraint. We show that the mechanism is asymptotically Bayesian incentive-compatible and hence asymptotically efficient. Surprisingly, without the ex post individual rationality constraint, the BayesianNash equilibrium strategy for the buyers is to bid more than their true value.