CSC304 Lecture 11
Mechanism Design w/ Money: Revelation principle; First price, second price, and ascending auctions; Revenue equivalence
CSC304 - Nisarg Shah 1
CSC304 Lecture 11 Mechanism Design w/ Money: Revelation principle; - - PowerPoint PPT Presentation
CSC304 Lecture 11 Mechanism Design w/ Money: Revelation principle; First price, second price, and ascending auctions; Revenue equivalence CSC304 - Nisarg Shah 1 Recap: Bayesian Framework
CSC304 - Nisarg Shah 1
CSC304 - Nisarg Shah 2
๐ฌ๐ ๐ฌ๐ ๐ฌ๐
๐๐ ๐๐ ๐๐
All distributions known to all agents Private value of ๐
๐๐ ๐๐ ๐๐
๐๐ ๐๐ ๐๐ All strategies known to all agents
CSC304 - Nisarg Shah 3
โข Interim utility of agent ๐ is
๐น ๐ค๐โผ๐ธ๐ ๐โ ๐ ๐ฃ๐ ๐ก1 ๐ค1 , โฆ , ๐ก๐ ๐ค๐
where utility ๐ฃ๐ is โvalue derived โ payment chargedโ
โข ิฆ
๐ก is a Bayes-Nash equilibrium (BNE) if ๐ก๐ is the best strategy for agent ๐ given ิฆ ๐กโ๐ (strategies of others)
rational players, so I can reason about what strategies they might use.
CSC304 - Nisarg Shah 4
โข Each agent ๐ privately submits a bid ๐๐ โข Agent ๐โ with the highest bid wins the item, pays ๐๐โ
โข Common prior: each has valuation drawn from ๐[0,1]
โข Proof on the board.
CSC304 - Nisarg Shah 5
CSC304 - Nisarg Shah 6
private values
โข Doesnโt mean agents will report their true values. โข Makes sense to ask โWould they, in equilibrium?โ
space
โข Suppose your value for an item is in [0,1], but the
mechanism asks you to either dive left or dive right.
โข Strategy ๐ก๐: 0,1 โ {๐๐๐๐ข, ๐ ๐๐โ๐ข} โข Truthfulness doesnโt make much sense. โข But we can still ask: What is the outcome in equilibrium?
CSC304 - Nisarg Shah 7
โข I donโt know what otherโs valuations are, only the
distributions theyโre drawn from.
โข But as long as they report their true values, in
expectation I would like to report my true value.
โข I know what othersโ values are, and for every possible
values they can have, I want to report my true values.
CSC304 - Nisarg Shah 8
โข If a mechanism implements an outcome in dominant
strategies, thereโs a direct revelation strategyproof mechanism implementing the same outcome.
โข If a mechanism implements an outcome as BNE, thereโs a
direct revelation BNIC mechanism implementing the same outcome.
CSC304 - Nisarg Shah 9
Player 1 : ๐ค1
Strategy s1 Player ๐ : ๐ค๐ Strategy s๐ Original Mechanism Outcome
New direct revelation truthful mechanism
CSC304 - Nisarg Shah 10
โข Sealed-bid 1st price auction โข 2 agents with valuations drawn from ๐[0,1] โข Each player halving his value was a BNE โข Not naturally BNIC (players donโt report value)
CSC304 - Nisarg Shah 11
CSC304 - Nisarg Shah 12
โข ๐น ๐ค๐โผ๐ 0,1
๐=1 ๐
๐โ1 ๐
โ max
๐
๐ค๐ =
๐โ1 ๐+1
(Exercise!)
โข ๐น ๐ค๐โผ๐ 0,1
๐=1 ๐ [2nd highest ๐ค๐] =
๐โ1 ๐+1
(Exercise!)
CSC304 - Nisarg Shah 13
type ๐ค๐
0 (e.g., โzero value for allโ โ zero payment);
โpath-connected support setsโ;
โข Charge the same expected payment to all agent types; โข Have the same expected total revenue.
CSC304 - Nisarg Shah 14
โข If two BNIC mechanisms always have the same allocation,
then they have the same E[payments] and E[revenue].
โข Very powerful as it applies to any pair of BNIC mechanism
โข Have the same allocation:
Item always goes to the agent with the highest valuation
โข Thus, also have the same revenue
CSC304 - Nisarg Shah 15
โข All agents + auctioneer meet in a room. โข Auctioneer starts the price at 0. โข All agents want the item, and have their hands raised. โข Auctioneer raise the price continuously. โข Agents drop out when price > value for them
โข Start price at a very high value. โข Keep decreasing the price until some agent agrees to buy.