Auctions used to sell houses often attract a diverse group of bidders, with realtors and speculators out for a bargain competing against buyers with a real interest in the house. Value asymmetries such as these necessitate careful consideration of the auction format as revenue equivalence cannot be expected to hold. From a theoretical viewpoint, Myerson's (1981) mechanism design approach has identified the seller's optimal choice. The proposed mechanism entails assigning credits to weaker bidders to promote competition and setting bidder-specific reserve prices. In practice, however, sellers often lack the detailed information needed to choose credits and reserve prices optimally, nor can they always discriminate among bidders. A more practical solution to the seller's problem is suggested by the "Amsterdam auction," where a premium is offered to encourage weak bidders to compete aggressively. This auction format, which has been used to sell houses in Amsterdam for centuries, treats all bidders the same and does not rely on detailed information about their value-distributions. In this paper, we consider premium auctions like the one in Amsterdam and demonstrate their revenue-generating virtues in asymmetric situations. We report the results of an experiment, which compares the standard first-price and English formats with two premium auctions in symmetric and asymmetric settings. The introduction of a premium leads weak bidders to set an endogenous reserve price for stronger rivals, with a dramatic effect on the sales price. Awarding a premium raises revenues, especially since Bertrand competition between weaker bidders virtually dissipates the premium to be paid.
