Behavior and learning in asymmetric independent private values auctions

This dissertation consists of three essays that employ experimental and computational methods to test theoretical models of asymmetric independent private value auctions.; The first essay reports simulation results from a model of adaptive learning in the context of an asymmetric auction. The model is designed to investigate the long-run effectiveness of adaptive learning and to explore the impact of different information environments ("social" vs. "individual" learning) and the degree of learning task complexity. Simulation results suggest that learning in different settings vary both in terms of convergence speed and payoffs. However, in all cases adaptive agents learn to use bidding functions that are very close to the Nash Equilibrium strategies.; In the second essay I experimentally test an auction model with asymmetries that can arise when a subset of symmetric bidders join to bid together. Possible applications include mergers, collusion and legal joint-bidding arrangements. The transformation produces a party (Strong) who has a more advantageous value distribution than the remaining (Weak) bidder(s). The essay reports important deviations from theoretical predictions. Thus, I find that the predicted negative impact on the seller's revenue is reduced as the Weak bidders tend to increase their bids relative to the symmetric benchmark. Economic efficiency actually improves. Finally, the incentives to bid jointly may be higher than predicted as the Strong party obtains higher per-member profits than the Weak bidders.; In the third essay a simple model of combinatorial auctions is developed and tested experimentally. A combinatorial auction allows bidding for packages of items and can improve economic efficiency when synergies among sale objects are high. However, it can be strategically "abused" by bidders when synergies are absent, hurting the efficiency. This trade-off is illustrated in a model with local and global bidders. "Free-riding" among local bidders induced by the combinatorial auction is an important force behind the trade-off. The experimental results lend strong support to the importance of this phenomenon. The observed changes in efficiency levels are mostly consistent with the theoretical predictions. Finally, regardless of the degree of synergies, the combinatorial auction generated significantly lower revenues both theoretically and in the experimental data.