Research Article
New Core-Selecting Payment Rules with Better Fairness and Incentive Properties
@INPROCEEDINGS{10.4108/eai.8-8-2015.2260633, author={Benjamin Lubin and Benedikt B\'{y}nz and Sven Seuken}, title={New Core-Selecting Payment Rules with Better Fairness and Incentive Properties}, proceedings={The Third Conference on Auctions, Market Mechanisms and Their Applications}, publisher={ACM}, proceedings_a={AMMA}, year={2015}, month={8}, keywords={combinatorial auctions payment rules core fairness}, doi={10.4108/eai.8-8-2015.2260633} }- Benjamin Lubin
Benedikt Bünz
Sven Seuken
Year: 2015
New Core-Selecting Payment Rules with Better Fairness and Incentive Properties
AMMA
ICST
DOI: 10.4108/eai.8-8-2015.2260633
Abstract
In this paper, we study payment rules for core-selecting combinatorial auctions (CAs). We consider the fact that in many real-world CAs, bidders are heterogeneous in size and value. To take this into account, we measure the "aggregate incentives" of a rule and introduce two "fairness measures" based on the Gini coefficient. As our benchmark, we study the "Quadratic" rule (Day and Cramton, 2012), which is the rule most commonly used in practice. We find that the Quadratic rule provides large incentives to deviate from truthful bidding and also produces an unfair distribution of payoffs, favoring larger bidders. Our research goal is to find a core-selecting payment rule that has good efficiency properties, but outperforms the Quadratic rule in terms of aggregate incentives and fairness.
We have designed 75 new rules built around different distance metrics, reference points, and bidder weightings. To find a core-selecting payment rule that strikes an optimal trade-off between all design objectives, we have evaluated 85 different rules using a new algorithm we have developed for finding approximate Bayes-Nash equilibria in CAs. Finally, we recommend using one of our new rules: the "Marginal-Economy-Fractional" rule with VCG as the reference point. We find that this rule generally performs as well as or even slightly better than the Quadratic rule in terms of efficiency and revenue, but significantly improves upon it in terms of aggregate incentives and fairness.