Auctions and Matching Recent Challenges Paul Milgrom MFI
- Slides: 9
Auctions and Matching: Recent Challenges Paul Milgrom MFI Conference on “Matching and Price Theory” May 7, 2011
Connection to Matching Kelso-Crawford… The simultaneous multiple round auction (SMRA), run through much of the world, works pretty well when Goods are substitutes Players are “near price takers” Main SMRA innovation: activity rule Curtails some strategic play Shortens the auction Almost universally adopted Complementarities problematic for auctions: present with a vengeance!
Upcoming Challenges in the US “Incentive Auctions” “Connect America” (Universal service fund) Shared difficulty: non-substitutes structure Dynamic reporting (see Segal) Structured message spaces
“Incentive Auctions” Repurposing Radio Spectrum Broadcasters and others may sell some spectrum, while Wireless broadband servicers simultaneously buy spectrum Challenges The broadcasters have isolated 6 MHz bands. Wireless broadband (LTE) needs more, optimally 2 x 20 MHz chunks of adjacent frequencies, and same frequency in every geographic area. FCC proposal: voluntary auctions, compulsory moves
Connect America Stated policy goal to “connect America” – provide coverage to more than 95% of the population. Providers have economies to serve particular geographic areas, and it is far cheaper to serve contiguous areas. What would an auction solution look like?
Vickrey auction? Problems Complexity for these applications Outcomes not in the core (very low or zero prices possible Unusual susceptibility to collusion
Vickrey Example Bidder Item A Item B Package AB 1 10* 1 10 2 1 10* 10 3 0 0 10 Three competitors for AB, each willing to pay 10 Vickrey prices are zero Outcome is not stable (not in the core).
CCA Auction Design Beginning with the UK in 2007, several European countries began to adopt versions of the dynamic “core-selecting auctions” (based in part on Day. Milgrom (2007) and in part on Ausubel-Cramton. Milgrom (2005)). New pricing rule always leads to outcomes in the core with respect to reported values. Coincides exactly with Vickrey when goods are substitutes for all bidders. Conjecture: has Budish’s near-strategy-proof property in a “reasonable” model (generalized core convergence)
End