1 / 56

The Clock-Proxy Auction: A Practical Combinatorial Auction Design

The Clock-Proxy Auction: A Practical Combinatorial Auction Design. Lawrence M. Ausubel, Peter Cramton, Paul Milgrom University of Maryland and Stanford University 8 October 2004. * Some of the methods discussed are subject to issued patents or pending applications. Introduction.

noah-fuentes
Download Presentation

The Clock-Proxy Auction: A Practical Combinatorial Auction Design

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. The Clock-Proxy Auction:A Practical Combinatorial Auction Design Lawrence M. Ausubel, Peter Cramton, Paul Milgrom University of Maryland and Stanford University 8 October 2004 * Some ofthe methods discussed are subject to issued patents or pending applications.

  2. Introduction • Many related (divisible) goods • Spectrum (location) • Electricity (duration, location, strike price, ancillary services) • Financial securities (duration) • Emissions (duration, type) • A practical combinatorial auction for FCC (and others) to replace simultaneous ascending auction (SAA)

  3. Introduction Clock Auction • Auctioneer names prices; bidders name only quantities • Price adjusted according to excess demand • Process repeated until market clears • No exposure problem (package auction)

  4. Introduction Proxy Auction • A procedure for package bidding • Bidders input their values into “proxy agents” • Proxy agents iteratively submit package bids, selecting best profit opportunity according to the inputted values • Auctioneer selects provisionally-winning bids according to revenue maximization • Process continues until the proxy agents have no new bids to submit

  5. Introduction Clock-Proxy Auction • A clock auction, followed by a “final round” consisting of a proxy auction • Bidders directly submit bids in clock auction phase • When clock phase concludes, bidders have a single opportunity to input proxy values • The proxy phase concludes the auction

  6. Introduction Clock-Proxy Auction • All bids are kept “live” throughout auction (no bid withdrawals) • Bids from clock phase are also treated as package bids in the proxy phase • All bids are treated as mutually exclusive (XOR) • Activity rules are maintained within clock phase and between clock and proxy phases

  7. Introduction Advantages of Clock-Proxy Auction • Clock phase • Simple for bidders • Provides essential price discovery • Proxy phase • Highly efficient • Competitive revenues • Little opportunity for collusion

  8. Clock Auction

  9. Simultaneous Clock Auction • Practical implementation of the fictitious “Walrasian auctioneer” • Auctioneer announces a price vector • Bidders respond by reporting quantity vectors • Price is adjusted according to excess demand • Process is repeated until the market clears

  10. Simultaneous Clock Auction • Strengths • Simple for bidders • Provides highly-usable price discovery • Yields similar outcome as SAA, but faster and fewer collusive opportunities • A package auction without complexity • Weaknesses • Limits prices to being linear • Therefore should not yield efficient outcomes

  11. Recent Clock Auctions (MDI) • EDF generation capacity (virtual power plants) • 13 quarterly auctions (Sep 2001 – present) • Electrabel generation (virtual power plants) • 4 quarterly auctions (Dec 2003 – present) • Ruhrgas gas release program • 2 annual auctions (2003 – present) • UK emissions trading scheme • World’s first greenhouse gas auction (Mar 2002) • GDF and Total gas release program • 2 auctions (Oct 2004) • FirstEnergy (Ohio) standard offer service • 1 annual auction (Nov 2004)

  12. Recent Clock Auctions (others) • New Jersey basic generation service • 3 annual auctions (2002 – present) • Texas electricity capacity • 12 quarterly auctions (Sep 2001 – present) • Austrian gas release program • 2 Annual Auctions (2003 – present) • Nuon generation capacity • One auction (July 2004)

  13. EDF Generation Capacity Auction MDI market design inc.

  14. Typical EDF Auction • Number of products • Two to five groups (baseload, peakload, etc.) • 20 products (various durations) • Number of bidders • 30 bidders • 15 winners • Duration • Eight to ten rounds (one day) • €200 million in value transacted in auction

  15. Electrabel VPP Capacity Auction MDI market design inc.

  16. Typical Electrabel Auction • Number of products • Two groups (baseload, peakload) • 20 products (various durations and start dates) • Number of bidders • 14 bidders • 7 winners • Duration • Seven rounds (one day) • €70 million in value transacted in auction

  17. Typical Ruhrgas Auction • Number of products • One (39 identical lots) • Number of bidders • 16 bidders • 7 winners • Duration: part of one day • €350 million in value transacted in auction

  18. Issues in Implementing Clock Auction • Bids need to be taken literally and need to be treated as binding contractual offers • PROBLEM: If bids need to be submitted unreasonably frequently or at unexpected intervals, bidders may miss making required submissions of bids • SOLUTION: Discrete bidding rounds • Avoiding “overshoot” • PROBLEM: Given discrete bidding rounds and need for a quick auction, bid increments need to be reasonably large, and price may overshoot the market-clearing price • SOLUTION: Intra-round bidding

  19. Round 6 P6 Closing Price: 1 Product – Dealing with Discreteness Price Overshoot Round 5 P5 Round 4 P4 Round 3 P3 Round 2 P2 Round 1 P1 MW Aggregate Demand Supply

  20. Round 6 Round 6 P6 Round 5 P5 Round 4 P4 Round 3 P3 Round 5 Round 2 P2 Round 1 P1 1 Product introducing intra-round bidding Price MW quantity bid by an individual

  21. Round 6 P6 Round 5 P5 Round 4 P4 Round 3 P3 Round 2 P2 Round 1 P1 1 product – Individual bids with intra-round bidding Price MW quantity bid by an individual

  22. Minimal Overshoot Round 6 P6 Closing Price Round 5 P5 Round 4 P4 Round 3 P3 Round 2 P2 Round 1 P1 1 product – Aggregate demand with intra-round bidding Price MW Aggregate Demand Supply

  23. Sample 1

  24. Sample 2

  25. Simultaneous Clock Auction Issue 2: Treatment of bids which would make aggregate demand < supply • Example: For a particular item, demand = supply, but the price of a complementary item increases. A bidder wishes to reduce its demand • Naive approach: Prevent the reduction • Example: For a particular item, demand > supply, but demand < supply at next increment • Naive approach: Ration the bidders

  26. Simultaneous Clock Auction Issue 2: Treatment of bids which would make aggregate demand < supply • Example: For a particular item, demand = supply, but the price of a complementary item increases. A bidder wishes to reduce its demand • Difficulty: Creates an exposure problem • Example: For a particular item, demand > supply, but demand < supply at next increment • Difficulty: Creates an exposure problem

  27. Simultaneous Clock Auction Issue 2: Treatment of bids which would make aggregate demand < supply • Example: For a particular item, demand = supply, but the price of a complementary item increases. A bidder wishes to reduce its demand • Our approach: Allow the reduction • Example: For a particular item, demand > supply, but demand < supply at next increment • Our approach: No rationing

  28. Simultaneous Clock Auction Issue 2: Treatment of bids which would make aggregate demand < supply • “Full Flexibility” (used in the EDF auctions; advocated here) • After each new price vector, bidders can arbitrarily reduce their previous quantities • Advantage • Makes clock auction into a combinatorial auction • No exposure problem! • Disadvantage • There may be significant undersell • Not a problem if it is followed by a proxy auction

  29. Simultaneous Clock Auction Issue 3: Activity rules • Prevent a bidder from hiding as a “snake in the grass” to conceal its true interests • Standard approaches: • No activity rule (laboratory experiments) • Monotonicity in quantities (SAA and clock auctions in practice)

  30. Simultaneous Clock Auction Issue 3: Activity rules • Revealed-preference activity rule (advocated here) • Compare times s and t (s < t), Prices: ps, ptDemands: xs, xt • At time s, xs is better than xt: • At time t, xt is better than xs : • Adding inequalities yields the RP activity rule:

  31. Simultaneous Clock Auction Issue 3: Activity rules • Revealed-preference activity rule (advocated here) • Bid placed at time t must satisfy (RP) with respect to its prior bids at all prior times s (s < t): • One can also apply a “relaxed” RP in proxy phase (with respect to bids in the clock phase)

  32. Proxy Auction

  33. Package Bidding • Package bidding often motivated by complements • Even without complements, package bidding may improve outcome by eliminating “demand reduction” • In SAA, bidders may have strong incentives to reduce demands in order to end auction at low prices

  34. Basic Ascending Package Auction • A set of items is offered for sale • A bid specifies a set of items and a corresponding bid amount • Bidding proceeds in a series of rounds • After each round, provisional winning bids are determined that maximize revenues • Auction ends after a round with no new bids • All bids are treated as mutually exclusive (XOR) • All bids are kept “live” throughout the auction

  35. Ascending Proxy Auction • Each bidder reports its values (and constraints) to a “proxy bidder” • Proxy bidder bids on behalf of the real bidder — iteratively submitting the allowable bid that, if accepted, would maximize the bidder’s payoff (evaluated according to its reported values) • Auction ends after a round with no new bids

  36. Example: Ascending Proxy Auction • Two items, A and B; bids must be integers • Bidder reports values of v(A) = 10, v(B) = 5, v(A,B) = 20 • Past high bids by this bidder (all “losing”) were: • b(A) = 4, b(B) = 3, b(A,B) = 15 • Next allowable bids are: • b(A) = 5 Yields profits of  = v(A) – b(A) = 10 – 5 = 5 • b(B) = 4 Yields profits of  = v(B) – b(B) = 5 – 4 = 1 • b(A,B) = 16 Yields profits of  = v(A,B) – b(A,B) = 20 – 16 = 4 • So the proxy bidder next places a bid of 5 on A

  37. Example: Ascending Proxy Auction • Two items, A and B; bids must be integers • Bidder reports values of v(A) = 10, v(B) = 5, v(A,B) = 20 • Past high bids by this bidder (all “losing”) were: • b(A) = 4, b(B) = 3, b(A,B) = 15 • Next allowable bids are: • b(A) = 5 Yields profits of  = v(A) – b(A) = 10 – 5 = 5 • b(B) = 4 Yields profits of  = v(B) – b(B) = 5 – 4 = 1 • b(A,B) = 16 Yields profits of  = v(A,B) – b(A,B) = 20 – 16 = 4 • Next allowable bids after that are: • b(A) = 6 Yields profits of  = v(A) – b(A) = 10 – 6 = 4 • b(B) = 4 Yields profits of  = v(B) – b(B) = 5 – 4 = 1 • b(A,B) = 16 Yields profits of  = v(A,B) – b(A,B) = 20 – 16 = 4 • So the proxy next bids 6 on A and/or 16 on {A,B}

  38. Outcomes in the Core • The coalitional form game is (L,w), where… • L denotes the set of players. • the seller is l = 0 • the other players are the bidders • w(S) denotes the value of coalition S: • If S excludes the seller, let w(S)=0 • If S includes the seller, let • The Core(L,w) is the set of all profit allocations that are feasible for the coalition of the whole and cannot be blocked by any coalition S

  39. Outcomes in the Core Theorem. (Ausubel and Milgrom 2002 , Parkes and Ungar 2000) The payoff vector resulting from the proxy auction is in the core relative to the reported preferences Interpretations: • Core outcome assures competitive revenues for seller • Core outcome assures allocative efficiency (ascending proxy auction is not subject to inefficient demand reduction)

  40. Case of Substitutes • If goods are substitutes, then Vickrey payoff profile is bidder-Pareto-optimal point in core • Outcome of the ascending proxy auction coincides with outcome of the Vickrey auction Vickrey Payoff Vector w(L)-w(L\2) Core Payoffs for 1 and 2 Bidder #2 Payoff v1+v2w(L)-w(L\12) Bidder #1 Payoff w(L)-w(L\1)

  41. Case of Non-Substitutes • If goods are not substitutes, then Vickrey payoff profile is not in core • Ascending proxy auction yields a different outcome from the Vickrey auction (one with higher revenues) Vickrey Payoff Vector w(L)-w(L\2) Bidder-Pareto-optimal payoffs Core Payoffs for 1 and 2 Bidder #2 Payoff v1+v2w(L)-w(L\12) Bidder #1 Payoff w(L)-w(L\1)

  42. Outcomes in the Core Theorem (Ausubel and Milgrom 2002). If  is a bidder-Pareto-optimal point in Core(L,w), then there exists a full information Nash equilibrium of the proxy auction with associated payoff vector . These equilibria may be obtained using strategies of the form: bid your true value minus a nonnegative constant on every package

  43. Proxy Auction Avoids Vickrey Problems • In Vickrey auction: • Adding a bidder can reduce revenues • Using a shill bidder can be profitable • Losing bidders can profitably collude

  44. Monotonicity and Revenue Issues • Example: Two identical items, A and B; three bidders • Bidder 1 values the pair only: v1(A,B) = $2 billion • Bidder 2 wants a single item only: v2(A) = $2 billion • Bidder 3 wants a single item only: v3(B) = $2 billion • The Vickrey auction awards each bidder his incremental value: • Bidders 2 and 3 each win one item • Social value with Bidder 2 = $4 billion; without Bidder 2 = $2 billion • Prices in the Vickrey auction equal zero! • The problem in this example is a failure of monotonicity: • Adding Bidder 3 reduces Vickrey revenues from $2 billion to zero • The Vickrey outcome lies outside the core • The proxy auction avoids this problem: Revenues = $2 billion

  45. The Loser Collusion Problem • Example: Two identical items, A and B; three bidders • Bidder 1 values the pair only: v1(A,B) = $2 billion • Bidder 2 wants a single item only: v2(A) = $0.5 billion • Bidder 3 wants a single item only: v3(B) = $0.5 billion • The losing Bidders 2 and 3 have a profitable joint deviation in the Vickrey auction: bidding $2 billion each • This converts it into the previous example • Bidders 2 and 3 each win one item at prices of zero • The Vickrey auction is unique in its vulnerability to collusion even among losing bidders • The proxy auction avoids this problem: Bidders 2 and 3 can overturn the outcome of Bidder 1 winning only by jointly bidding $2 billion

  46. The Shill Bidding Problem • Example: Two identical items, A and B; two bidders • Bidder 1 values the pair only: v1(A,B) = $2 billion • Bidder 2 has v2(A) = $0.5 billion; v2(A,B) = $1 billion • The losing Bidder 2 can set up a bidder under a false name (“shill bidder”). Each of Bidder 2 and the shill Bidder 3 can bid $2 billion each • This again converts it into the first example • Bidder 2 wins two items and pays zero! • The Vickrey auction is vulnerable to shill bidding

  47. Clock-Proxy Auction

  48. Clock-Proxy Auction • A simultaneous clock auction is conducted, with a revealed-preference activity rule imposed on bidders, until (approximate) clearing is attained • A proxy auction is conducted as a “final round” • Bids submitted by proxy agents are restricted to satisfy a relaxed revealed-preference activity rule based on competitive conditions • Bids from clock phase are also treated as “live” package bids in proxy phase • All package bids (clock and proxy) are treated as mutually exclusive, and auctioneer selects as provisionally-winning the bids that maximize revenues

  49. Relaxed Revealed Preference Activity Rule • Let s be a time in clock phase and t a time in proxy phase • Package S is bid on at time s and T is bid on at time t • Ps(S) and Ps(T) package prices of S and T at time s • Pt(S) and Pt(T) package prices of S and T at time t • At every time t in the proxy phase, the bidder can bid on the package T only if (RRP) is satisfied for every package S bid at time s in the clock phase • (RRP) [Pt(S) – Ps(S)] Pt(T) – Ps(T) •  > 1 is parameter (closer to 1 if more competitive environment) • For  = 1, price of S increased more than price of T; otherwise S would be more profitable than T. • Alternatively, state RRP as a constrain on valuations reported to proxy:

  50. Why Not Use the Proxy Auction Only? • Clock auction phase yields price discovery • Feedback of linear prices is extremely useful to bidders • Clock phase makes bidding in the proxy phase vastly simpler • Focus decision on what is relevant • See what you don't need to consider • See what looks like good possibilities

More Related