Abstract
We analyze an e-market design that allows multiple market segments to be served simultaneously with a single generalized combinatorial auction. The mechanism uses rule-based bids designed to accommodate various kinds of bidders, such as those more sensitive to price or those more restricted in their requirements. We demonstrate experimentally—using agent-based simulation of the actual market for television advertising slots—that the rule-based approach effectively handles the wide range of market segments, while maintaining buyer and seller surplus and efficiently allocating goods.