On Maximizing Revenue for Multi-Item Auctions
summary
Proves a concrete constant-factor approximation for revenue in independent multi-item, multi-buyer auctions by decomposing the problem via an N-to-one reduction into disjoint one-buyer K-item problems, then designing a deterministic truthful mechanism called beta-bundling (choose per-buyer beta thresholds, sell separately or bundle with surcharge). Shows DBG mechanism is constant-competitive with optimal revenue, bounds the “dumping tax” by second-price revenue, and extends the bound to BIC vs DIC.
AI-written summary · about