Presentation on theme: "Bayesian games and their use in auctions"— Presentation transcript:
1Bayesian games and their use in auctions Adapted from notes by Vincent Conitzer
2What is mechanism design? In mechanism design, we get to design the game (or mechanism)e.g. the rules of the auction, marketplace, election, …Goal is to obtain good outcomes when agents behave strategically (game-theoretically)Mechanism design is often considered part of game theory2007 Nobel Prize in Economics!Before we get to mechanism design, first we need to know how to evaluate mechanisms
3Which auction generates more revenue? Each bid depends onbidder’s true valuation for the item (utility = valuation - payment),bidder’s beliefs over what others will bid (→ game theory),and... the auction mechanism usedIn a first-price auction, it does not make sense to bid your true valuationEven if you win, your utility will be 0…In a second-price auction, it always makes sense to bid your true valuationbid 1: $10a likely outcome for the first-price mechanisma likely outcome for the second-price mechanismbid 1: $5bid 2: $5bid 2: $4bid 3: $1bid 3: $1Are there other auctions that perform better? How do we know when we have found the best one?
4Collusion in the Vickrey auction Example: two colluding biddersv1 = first colluder’s true valuationv2 = second colluder’s true valuationprice colluder 1 would pay when colluders bid truthfullygains to be distributed among colludersb = highest bid among other biddersprice colluder 1 would pay if colluder 2 does not bid
5The term “Bayesian”"Bayesian" refers to the mathematician and theologian Thomas Bayes (1702–1761), who provided the first mathematical treatment of a non-trivial problem of Bayesian inference.Bayesian probability is one of the different interpretations of the concept of probability and belongs to the category of evidential probabilities
6The probability of H and D (joint probability) can be thought of in two ways: P(H,D) = P(H|D) * P(D)= P(D|H)*P(H)
8Now, driving is not independent from gender Now, driving is not independent from gender. Given someone is a girl, what is theprobability she doesn’t drive?Given someone doesn’t drive, what is the probability she is female?
10Bayes theorem: The posterior probability is proportional to the likelihood of the observed data, multiplied by the prior probability. ThusP(H|D) = P(D|H)*P(H)/P(D)P(H) is the prior probability of H: the probability that H is correct before the data D are seen.P(D|H) is the conditional probability of seeing the data D given that the hypothesis H is true. This conditional probability is called the likelihood.P(D) is the probability of D.P(H|D) is the posterior probability: the probability that H is true, given the data and previous state of beliefs about H.
11ExampleMarie is getting married tomorrow. In recent years, it has rained only 5 days each year. Unfortunately, the weatherman has predicted rain for tomorrow. When it actually rains, the weatherman correctly forecasts rain 90% of the time. When it doesn't rain, he incorrectly forecasts rain 5% of the time. What is the probability that it will rain on the day of Marie's wedding?P(rain) = 5/365P(not raining) = 360/365P(weatherman predicting rain|rain) = .9P(weatherman predicting rain|does NOT rain) = .05
12What is the probability of rain, given the weatherman predicts rain? P( rain | predict rain) = P(rain) P( predict rain| rain ) / P(predict rain)P(predict rain) = [P( rain)* P( predict rain| rain ) +P( no rain)* P( predict rain| no rain ) ]P( rain | predict rain) =(0.014)(0.9) / [ (0.014)(0.9) + (0.986)(0.05)P( rain | predict rain)] =
13Bayesian gamesWhat if we didn’t even know what game is being played? (Note that in imperfect information games, we didn’t know prior moves, but we at least knew the payoffs)On the exam, the “gift” question – both players knew whether or not a gift had been given, but suppose they didn’t have that knowledge?We say the “type” of agent determines the game we are playing
14Expected Utility Ex post: utility based on all agent’s actual types Ex interim: agent knows his own type, but not that of othersEx ante: the agent does not know anybody’s type
15Bayesian games L R row player type 1 U D L R U row player type 2 D In a Bayesian game a player’s utility depends on that player’s type as well as the actions taken in the gameNotation: θi is player i’s type, drawn according to some distribution from set of types ΘiIn this example, each player knows/learns its own type, not those of the others, before choosing actionPure strategy si is a mapping from Θi to Ai (where Ai is i’s set of actions)In general players can also receive signals about other players’ utilities. Assume all know the probabilities of being in each game.LRcolumn playerrow playertype 1U2,00,22,20,33,01,1D(.3)(.1)LRUrow playertype 22,20,01,12,10,01,2(.2)(.4)D
16Induced normal form (based on probabilities – no types are known) LLLRRLRRUU2, 11, .71, 1.20,.9UD.8, .21, 1.1.4, 1.6, 1.9DU1.5, 1.4.5, 1.11.7, .4.7, .1DD.3, .6.5, 1.51.1, 0.21.3, 1.1If we had a NE in this induced normal form, it would be called a Bayes NE.
17Induced normal form (based on probabilities given player 1 knows his type is type 1) LLLRRLRRUU2, 0.51.5,.75.5, 20, 2.25UDDU.25, 1.752.25, 01.75,.25DDSince player 1 knows he is in top games, in left game ¾ time, right ¼.
18Could rewrite with only two rows LLLRRLRRU2, 0.51.5,.75.5, 20, 2.25D.25, 1.752.25, 01.75,.25Since player 1 knows he is in top games, in left game ¾ time, right ¼.Meaningless to find Nash equilibrium,as payoffs aren’t common knowledge
19Bayes-Nash equilibrium A profile of strategies is a Bayes-Nash equilibrium if it is a Nash equilibrium for the normal form of the gameMinor caveat: each type should have >0 probabilityAlternative definition: for every i, for every type θi, for every alternative action ai, we must have:Σθ-i P(θ-i) ui(θi, σi(θi), σ-i(θ-i)) ≥Σθ-i P(θ-i) ui(θi, ai, σ-i(θ-i))
20First-price sealed-bid auction BNE Suppose every bidder (independently) draws a valuation from [0, 1]What is a Bayes-Nash equilibrium for this?Say a bidder with value vi bids vi(n-1)/nClaim: this is an equilibrium!Proof of equilibrium: suppose all others use strategyFor a bid b < (n-1)/n, the probability of winning is when the other n-1 bidders bid less than b. For a bidder to bid b, his valuation would be vi = bn/(n-1) Since he actually bids vi(n-1)/n = bn/(n-1)(n-1)/n = bBecause valuations are uniformly distributed (and the range is 0-1) the probability of anyone bidding b or less, is (bn/(n-1))And the probability of all n-1 others bidding b or less is (bn/(n-1))n-1
21First-price sealed-bid auction BNE For a bid b < (n-1)/n, the probability of winning is (bn/(n-1))n-1, so the expected value is (vi-b)(bn/(n-1))n-1Derivative w.r.t. b is (using product rule)- (bn/(n-1))n-1 + (vi-b)(n-1)bn-2(n/(n-1))n-1 which should equal zero- (bn/(n-1))n-1 + (vi-b)(n-1)bn-2(n/(n-1))n-1 = 0Dividing both sizes by bn-2(n/(n-1))n-1Implies -b + (vi-b)(n-1) = 0, which solves to b = vi(n-1)/n
22Analyzing the expected revenue of the first-price auction First-price auction: probability of there not being a bid higher than b is (bn/(n-1))n (for b < (n-1)/n)This is the cumulative density function of the highest bidProbability density function is the derivative, that is, it is nbn-1(n/(n-1))nExpected value of highest bid isn(n/(n-1))n∫(n-1)/nbndb = (n-1)/(n+1)Upper bound of integral chosen as no bid will be higher
23Revenue equivalence theorem Suppose valuations for the single item are drawn i.i.d. from a continuous distribution over [L, H] (with no “gaps”), and agents are risk-neutralThen, any two auction mechanisms thatin equilibrium always allocate the item to the bidder with the highest valuation, andgive an agent with valuation L an expected utility of 0,will lead to the same expected revenue for the auctioneer
24What if bidders are not risk-neutral? Behavior in second-price/English/Japanese does not change, but behavior in first-price/Dutch doesRisk averse: first price/Dutch will get higher expected revenue than second price/Japanese/EnglishRisk seeking: second price/Japanese/English will get higher expected revenue than first price/Dutch
25With common valuations E.g. bidding on drilling rights for an oil fieldEach bidder i has its own geologists who do tests, based on which the bidder assesses an expected value vi of the fieldIf you win, it is probably because the other bidders’ geologists’ tests turned out worse, and the oil field is not actually worth as much as you thoughtThe so-called winner’s curseSo if you have common valuation, what would you bid in a second price auction?
26With common valuations E.g. bidding on drilling rights for an oil fieldEach bidder i has its own geologists who do tests, based on which the bidder assesses an expected value vi of the fieldIf you win, it is probably because the other bidders’ geologists’ tests turned out worse, and the oil field is not actually worth as much as you thoughtThe so-called winner’s curseHence, bidding vi is no longer a dominant strategy in the second-price auctionIn English and Japanese auctions, you can update your valuation based on other agents’ bids, so no longer equivalent to second-priceIn these settings, English (or Japanese) > second-price > first-price/Dutch in terms of revenue
27ProblemYou and your siblings have been left with an estate consisting of many itemsThe value of each item is correlated (partially common and partially private)How can you divide the property so that the division is fair?
28Auctions without a seller faculty from a certain department in a university may want to decide who get to use a limited number of seminar rooms on a certain day, so that the department as a whole benefits the most;a group of housemates who collectively own a car may want to decide who gets to use it on a particular weekend, so that the one who needs it the most gets to use it.Agents are self interested and may lie about their valuation.
29CautionWe are programmed to think that auctions are about earning money. It may be that we are only trying to distribute goods fairly.In the case of dividing the estate, if there is no cost, people could be greedy or overstate their utility (given an inefficient allocation, right?).Children’s book: Instead of charging people for items, everything was given away. The logic was that people who made baseball bats loved to do it, so they were happy to give them away to someone who would enjoy it. Then, they could make more.I remember thinking this was a great idea.
30no deficit: it is reasonable to assume that a mechanism shall not subsidize any allocation it comes up with, i.e. the total payments made by the agents to the center must always be non-negativea stronger version of the no deficit property called strong budget balance, which says the total payments made by the agents to the center must be zero (useful property if jointly owned)strategy proofness means `truthful in dominant strategy'.
31Redistribute as much of the VCG revenue as possible back to the agents VCG mechanism fails to have the strong budget balance property (as the auctioneer makes money), it is in fact impossible to find a mechanism that has this property and at the same time is truthful and efficient in dominant strategySolutions?Redistribute as much of the VCG revenue as possible back to the agentsSacrifice truthfulnessSacrifice efficiency (social optimality)
32Vickrey auction without a seller ) = 2v() = 4v() = 3pays 3(money wasted!)But if there is no cost, people won’t be honest about their valuation
33Can we redistribute the payment? Idea: give everyone 1/n of the paymentv() = 2v() = 4v() = 3receives 1receives 1pays 3receives 1not strategy-proof: A loser (bidding higher) increases his redistribution paymentHe gets 1/n of the “value” as determined by the auction- so he benefits by increasing the selling price
34Incentive compatible redistribution [Bailey 97, Porter et al Incentive compatible redistribution [Bailey 97, Porter et al. 04, Cavallo 06]Idea: give everyone 1/n of second-highest other bidsv() = 2v() = 4v() = 3receives 1Who pays what?Strategy-proofYour redistribution does not depend on your bid;incentives are the same as in Vickrey
35Incentive compatible redistribution [Bailey 97, Porter et al Incentive compatible redistribution [Bailey 97, Porter et al. 04, Cavallo 06]Idea: give everyone 1/n of second-highest other bidsv() = 2v() = 4v() = 3receives 1pays 3receives 2/3receives 2/32/3 wasted (22%)Strategy-proofYour redistribution does not depend on your bid;incentives are the same as in Vickrey
36Bailey-Cavallo mechanism… Bids: V1≥V2≥V3≥... ≥Vn≥0First run Vickrey auctionPayment is V2 by bidder 1First two bidders receive V3/nRemaining bidders receive V2/nTotal redistributed: 2V3/n+(n- 2)V2/nSo why can’t we give it all to the remaining bidders?R1 = V3/nR2 = V3/nR3 = V2/nR4 = V2/n...Rn-1= V2/nRn = V2/n
37Another redistribution mechanism Bids: V1≥V2≥V3≥V4≥... ≥Vn≥0First run VickreyRedistribution:Receive 1/(n-2) * second- highest other bid,- 2/[(n-2)(n-3)] third-highest other bidTotal redistributed:V2-6V4/[(n-2)(n-3)]R1 = V3/(n-2) - 2/[(n-2)(n-3)]V4R2 = V3/(n-2) - 2/[(n-2)(n-3)]V4R3 = V2/(n-2) - 2/[(n-2)(n-3)]V4R4 = V2/(n-2) - 2/[(n-2)(n-3)]V3...Rn-1= V2/(n-2) - 2/[(n-2)(n-3)]V3Rn = V2/(n-2) - 2/[(n-2)(n-3)]V3Key point is their return is never a function of their bidIdea pursued further in Guo & Conitzer 07 / Moulin 07