Multi-object auctions with package bidding: An experimental comparison of Vickrey and iBEA

The use of package auctions for complex resource allocation has been rapidly increasing in recent years. In this paper, we study two package auction mechanisms in a laboratory setting, a sealed bid Vickrey auction and an ascending version of Vickrey, the iBEA auction. Unlike the single-unit Vickrey auction, where bidders tend to overbid in the laboratory, most of our bidders either underbid or bid their true values. Furthermore, at the aggregate level, while the Vickrey auction generates significantly higher revenue than does iBEA, the iBEA auction generates significantly higher bidder profit and efficiency. Additionally, a significantly larger proportion of iBEA auctions achieves 100% efficiency than does the Vickrey auction. We also find that human bidders learn from their robot opponents when the robot strategies are (myopic) best responses.