An Analysis of Linear Prices in Iterative Combinatorial Auctions

Ask prices are the most essential form of information feedback in Iterative Combinatorial Auctions (ICA) and vital in achieving an efficient auction outcome. Different pricing concepts have been analyzed in the literature, including linear, non-linear and discriminative non-linear prices. Linear prices, i.e. per-item prices, are intuitive and easy to understand for bidders, however, under certain conditions they can hinder auction efficiency. In this paper we analyze ways how approximate linear prices are calculated, identify problems of existing approaches and suggest alternative rules for calculating linear prices and stopping the auction. We use computer simulations as a tool to benchmark different auction designs, and compare their results based on allocative efficiency. Keyword: combinatorial auction, efficiency, linear prices, discrete event simulation