Characterizing strategy-proof, revenue monotone allocation rules in auction mechanisms

Taiki Todo, Atsushi Iwasaki, Makoto Yokoo

研究成果: Contribution to journalArticle査読

抄録

An auction mechanism consists of an allocation rule and a payment rule. There have been several studies on characterizing strategy-proof allocation rules, i.e., if the allocation rule satisfies a condition called weak-monotonicity, an appropriate payment rule is guaranteed to exist. One desirable property that an auction mechanism should satisfy is revenue monotonicity, i.e., a seller's revenue is guaranteed to weakly increase as the number of bidders grows. In this paper, we first identify a simple condition called summation-monotonicity for characterizing strategy-proof and revenue monotone allocation rules. To the best of our knowledge, this is the first attempt to characterize revenue monotone allocation rules. Based on this characterization, we also examine the connections between revenue monotonicity and false-name-proofness, which means a bidder cannot increase his utility by submittingmultiple bids under fictitious names. These two concepts look quite similar but their interaction is complicated. In a single-item auction, we show that they are basically equivalent; a mechanism is false-name-proof if and only if it is strategy-proof and revenue monotone. On the other hand, we show these two conditions cannot coexist in combinatorial auctions; under some minor conditions, there exists no combinatorial auction mechanism that is simultaneously revenue monotone and false-name-proof.

本文言語英語
ページ(範囲)86-96
ページ数11
ジャーナルTransactions of the Japanese Society for Artificial Intelligence
26
1
DOI
出版ステータス出版済み - 2011

All Science Journal Classification (ASJC) codes

  • ソフトウェア
  • 人工知能

引用スタイル