An Experimental Comparison of Linear and Nonlinear Price Combinatorial Auctions

Abstract

Combinatorial auctions are used for the efficient allocation of heterogeneous goods and services. They require appropriate software platforms providing automated winner determination and decision support for bidders. Several promising ascending combinatorial auction formats have been developed throughout the past few years based on primal-dual algorithms and linear programming theory. The Ascending Proxy Auction (Ausubel and Milgrom 2006a) and iBundle (Parkes and Ungar 2000) result in Vickrey payoffs when the coalitional value function satisfies buyer submodularity conditions and bidders bid best-response. These auction formats are based on non-linear and personalized ask prices. In addition, there are a number of designs with linear prices that have performed well in experiments (Bichler et al. 2009, Kwasnica et al. 2005. In this paper, we provide the results of lab experiments testing these different auction formats in the same setting. We analyze aggregate metrics, such as efficiency and auctioneer revenue for small and medium-sized value models. In addition, we provide a detailled analysis not only of aggregate performance metrics, but of individual bidding behavior under alternative combinatorial auction formats.

Ascending Combinatorial Auctions with Allocation Constraints: On Game Theoretical and Computational Properties of Generic Pricing Rules

“…FCA DL can reduce the number of auction rounds, which is a considerable problem of iBundle as shown by Scheffel et al (2011) andSchneider et al (2010). The reason is that dead bids in iBundle, which will never be part of the winning allocation, are skipped and prices increase faster.…”

“…These pricing rules are independent of the allocation rules, and laboratory experiments with respective auction formats yielded high levels of efficiency (Adomavicius et al 2012). They are very generic and can be considered a fundamental contribution in the emerging Information Systems literature on decision support in smart markets (Xia et al 2004, Bapna et al 2007, Guo et al 2007, Bichler et al 2009, Scheffel et al 2011 and in the general literature on CAs.…”

On the robustness of non-linear personalized price combinatorial auctions

“…Bidders were significantly more likely to bid on packages with a high temporary profit, but did not follow the pure straightforward strategy. More recently, Scheffel et al (2009) conducted experiments comparing iBundle, ALPS, Combinatorial Clock, and the VCG auction. Again, bidders in the iBundle auction did not follow the straightforward strategy, even though they were provided with a decision support tool that helped them select their demand set.…”

“…In particular, straightforward bidding might not hold in practical settings where bidders have bounded rationality, given that bidders don't know, whether the submodularity condition holds, and there is a huge number of bundles a bidder has to deal with. Recent experimental work has actually shown that bidders did not follow a pure best-response strategy, even in simple settings with only a few items (Scheffel et al 2009). Therefore, it is important to understand their performance in case of non-straightforward bidding strategies, when bidders either cannot follow such a strategy for computational or cognitive reasons, or deliberately choose another strategy.…”

Toward Understanding the Dynamics of Bidder Behavior in Continuous Combinatorial Auctions: Agent-Based Simulation Approach

“…While there has been research on a number of topics in this area -e.g., winner determination in combinatorial auctions, combinatorial auction designs, practicality of these designs for online marketplaces, and comparison of different auction mechanisms -the important issues related to bidder behavior in these auctions have been largely underexplored [2,3]. The main difficulty is that it is not possible to control for bidder behavior in experimental studies, which makes it hard to address a number of important and interesting research questions, for example, understanding how bidder behavior changes when facing different types of competition, and how these changes affect auction outcomes.…”