Ex Ante Information and the Design of Keyword Auctions

Abstract

This paper discusses a class of auctions, weighted unit-price auctions (WUPAs), which capture key features of keyword auctions, a novel mechanism behind the multi-billion-dollar keyword advertising industry. We analyze the equilibrium bidding strategy in the WUPA class and study its two main design parameters - weighting factors and minimum bids - both of which make use of the auctioneer's ex-ante information on bidders' ability to generate outcomes. Our results indicate equilibrium bidding functions in WUPAs may have kinks and jumps. WUPAs can be efficient when an auctioneer weights unit-price bids by bidders' expected yield and imposes the same minimum score (but not the same minimum bid-price) across all bidders. Optimally weighted WUPAs can generate more revenue than generalized first-price auctions, and optimal minimum bids generally differ from those prescribed in the mechanism design literature.

Keywords: keyword advertising, weighted unit-price auctions, ex-ante information, score rule, Google and Yahoo