TY - GEN
T1 - Lookahead Auctions with Pooling
AU - Feldman, Michal
AU - Gravin, Nick
AU - Tang, Zhihao Gavin
AU - Wald, Almog
N1 - Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2022
Y1 - 2022
N2 - A Lookahead Auction (LA), introduced by Ronen, is an auction format for the sale of a single item among multiple buyers, which is considered simpler and more fair than the optimal auction. Indeed, it anonymously selects a provisional winner by a symmetric ascending-price process, and only then uses a personalized posted price. A LA auction extracts at least 1/2 of the optimal revenue, even under a correlated value distribution. This bound is tight, even for 2 buyers with independent values. We introduce a natural extension of LA, called lookahead with pooling (LAP). A LAP auction proceeds as LA, with one difference: it allows the seller to pool together a range of values during the ascending-price stage, and treat them the same; thus, it preserves the simplicity and fairness of LA. Our main result is that this simple pooling operation improves the revenue guarantees for independent buyers from 1/2 to 4/7 of the optimal revenue. We also give a complementary negative result, showing that for arbitrary correlated priors LAP cannot do better than 1/2 approximation.
AB - A Lookahead Auction (LA), introduced by Ronen, is an auction format for the sale of a single item among multiple buyers, which is considered simpler and more fair than the optimal auction. Indeed, it anonymously selects a provisional winner by a symmetric ascending-price process, and only then uses a personalized posted price. A LA auction extracts at least 1/2 of the optimal revenue, even under a correlated value distribution. This bound is tight, even for 2 buyers with independent values. We introduce a natural extension of LA, called lookahead with pooling (LAP). A LAP auction proceeds as LA, with one difference: it allows the seller to pool together a range of values during the ascending-price stage, and treat them the same; thus, it preserves the simplicity and fairness of LA. Our main result is that this simple pooling operation improves the revenue guarantees for independent buyers from 1/2 to 4/7 of the optimal revenue. We also give a complementary negative result, showing that for arbitrary correlated priors LAP cannot do better than 1/2 approximation.
KW - Auction Design
KW - Lookahead Auctions
KW - Revenue Maximization
UR - http://www.scopus.com/inward/record.url?scp=85138811160&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-15714-1_4
DO - 10.1007/978-3-031-15714-1_4
M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???
AN - SCOPUS:85138811160
SN - 9783031157134
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 60
EP - 77
BT - Algorithmic Game Theory - 15th International Symposium, SAGT 2022, Proceedings
A2 - Kanellopoulos, Panagiotis
A2 - Kyropoulou, Maria
A2 - Voudouris, Alexandros
PB - Springer Science and Business Media Deutschland GmbH
T2 - 15th International Symposium on Algorithmic Game Theory, SAGT 2022
Y2 - 12 September 2022 through 15 September 2022
ER -