A framework for computing the outcome of proxied combinatorial auctions

Proxy bidding has been proposed for combinatorial auctions as a means to speed up the auctions, to simplify the user interface, and to limit strategic behavior. The only previously known solution method for proxy bidding in combinatorial auctions requires the auctioneer to run the auction with myopic bidders to determine the outcome. In this paper we present a radically different approach that computes the bidders' allocation of their attention across the bundles only at the points at which they change their bidding patterns. This algorithm has several advantages over alternatives, including that it computes exact solutions and is invariant to the magnitude of the bids. We present a general framework and apply it to Ausubel and Milgrom's APA mechanism and our own simple combinatorial proxy auction. We present an example in which the approach is applied to a multistage proxy auction, and report on some preliminary computational results.