One-Sided and Two-Sided Auction Design

Michael Parente

Advisor: .

Committee Members: Daniel Houser, Stephen Rassenti, David Porter

Truland Building, 400R
November 26, 2007, 07:00 PM to 07:00 PM

Abstract:

This dissertation introduces and tests a method for transforming any one-sided auction into a two-sided auction by allowing sellers to participate as if they were buyers for their own items. The institution proposed, an ascending auction for multiple units, is similar in design to the one-sided auction that the FCC uses to sell spectra licenses and the design could readily be extended to allow for package bidding from both buyers and sellers. I report three findings: 1) In environments with large traders, a two-sided auction discourages strategic behavior--both demand and supply reduction--and achieves higher efficiencies than an isomorphically equivalent one-sided auction. 2) With experience, sellers behave consistent with theory and make the same supply decisions regardless of whether their decisions are "framed" as buying back their own items or as making incremental additions to supply. Furthermore, I find no evidence that sellers in the position of buying back their own items can push prices in their favor (above the competitive equilibrium) any more than when sellers make supply decisions directly. 3) Inconsistent with current theory, sellers reduce supply less than buyers reduce demand when placed in roles that are strategically isomorphic (via ascending and descending versions of the clock prices). To confirm this result, further experimentation might test whether a descending uniform price auction in which only sellers are active would be more allocatively efficient than an equivalent ascending uniform price auction with only buyers.