Practical Auction Design Peter Cramton Professor of Economics, - PowerPoint PPT Presentation

Practical Auction Design Peter Cramton Professor of Economics, University of Maryland Chairman, Market Design Inc. President, Criterion Auctions 14 July 2006

Introduction • Auction design – Government perspective (design) – Bidder perspective (strategy) • Based on my experience – Researching auctions – Advising governments (a dozen) – Advising bidders (more than two dozen)

Advising bidders

Advising bidders: Typical project • Comment on auction design (regulatory) • First strategy meeting – Objectives – Strategic issues • Develop bid tracking tool • Second strategy meeting – Development of strategy – Several mock auctions • Daily auction advice during auction – Comment on auction strategy and end-of-day report

strategic issues Fundamental

Exercise • Two bidders (L & R), two items (A &B) – L need both (value = 2 times birth month) – R needs one (value = birth month) • Three auction formats – First price – Second price – Simultaneous ascending auction

Exposure problem

Exposure problem • With complements, bidding on individual lots is risky – Bidder must “go for it” or drop out early – Outcome is often inefficient – Experiments sometimes get high revenues • Exposure problem eliminated with package bids

Exposure problem exercise: Optimal strategy in SAA L's Profit -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 L's value 2 4 6 8 10 12 14 16 18 20 22 24 L's birth month 1 2 3 4 5 6 7 8 9 10 11 12 1 3 2 5 3 7 4 9 5 11 Bidder R's Cost to 6 13 birth eliminate 7 15 month bidder R 8 17 9 19 10 21 11 23 12 25 Expected Cost 14

Exposure problem exercise TU Berlin, 14 July 2006 First Price Optimal Second Price Bidder L Bidder R Near Opt Bidder L Bidder R VL/2 A B VR A B Bad Strat VL/2 A B VR A B 1 0 0 1 0.5 0 1 1 1 1 1 0 3 3 3 3 4 2 3 4 3 3 5 3 6 6 6 10 8 0 6 8 8 10 0 10 8 8 8 10 0 10 8 10 10 10 0 11 9 9 9 11 10 0 9 10 10 11 0 11 11 10 10 11 5.5 0 11 13 11 11 11 0 11 8 8 11 10 10

Exposure problem exercise AGATE, 4 July 2006 First Price Optimal Second Price Bidder L Bidder R Near Opt Bidder L Bidder R VL/2 A B VR A B Bad Strat VL/2 A B VR A B 1 0.01 0.01 1 0 0.99 1 3 0 1 0 1 2 0.9 0.9 3 2.5 0 2 1 1 3 3 0 3 2.99 2.99 4 0.5 3 3 2.99 2.99 4 0.5 3.5 4 3 3 5 2 2 4 3.95 3.95 5 3 3 4 2 3 5 3.5 0 4 4 4 5 0 5 4 3 3 5 2 3 4 4 4 5 5 3 5 5 4 6 0.6 1.6 5 10 10 6 0 6 7 7 7 7 0 4 7 7 7 7 0 7 8 4 4 8 0 6 8 8 8 8 0 8 9 1.5 1.5 8 6.01 0 9 3 3 8 8 0 12 10 10 8 0 7 12 12 12 8 8 0 9 7 0 9 0 9 10 7 3 10 4 10 12 4.07 8.05 12 12 6 12 0 8 12 0 12

Exposure problem exercise Ofcom, Summer 2006 First Price Optimal Second Price Bidder L Bidder R Near Opt Bidder L Bidder R VL A B VR A B Bad Strat VL A B VR A B 1 0.45 0.45 1 0 0.95 1 0.45 0.45 1 0 1 3 0 0 1 0 0.75 3 6 6 1 0 1 4 3 3 3 0 3 4 4 0 3 0 3 4 2.45 2.45 4 3.5 4 4 2.97 2.97 4 4 3.5 4 0 0 4 2.5 0 4 0 0 4 4 0 6 3 3 7 7 0 6 4 4 7 7 0 6 0 0 8 6.01 0 6 0 0 8 8.01 2.01 8 7 7 9 8.11 0 8 16 16 9 9 5 8 7 7 9 0 5 8 7 7 9 8 0 9 15 15 11 10 1 9 18 18 11 11 0 9 0 0 11 0 10 9 0 0 11 0 11 9 0 0 11 0 10 9 0 0 11 0 11 10 9.2 7.2 10 10.1 10 10 0 0 10 10 10 11 12 1 11 11 11

Objectives

Government goals • Competition in post-auction market • Efficiency • Revenue

Bidder preferences

Bidder preferences • Substitutes • Complements • Heterogeneity • Externalities

ascending auction Simultaneous

Auction rules • Simultaneous – All lots at the same time • Ascending – Can raise bid on any lot • Stopping rule – All lots open until no bids on any lot • Activity rule – Must be activity to maintain eligibility

Strategy in SAA

Strategy in SAA • Translating valuations into strategy • Auction as a negotiation • Elements of a good strategy • Retaliation • Managing eligibility • Auction opening • Stage transitions • Auction closing

Auction as a Negotiation • Learn what competitors need • Learn how costly it is to ask for more • Ask for more at the right time in the right place • Manage eligibility

Elements of a Good Strategy • Conservative (don’t leave money on table) • Flexible (responsive to others) • Clear • Nice, but firm – Begin with a cooperative position – “Quick to punish, quick to forgive”

Retaliation

Retaliatory Bidding • Stake a claim – Bid early on licenses of primary interest – Consistently retake license whenever bumped • Punish intruders – Punish by bumping intruder from a license it holds – Works well against rivals with primary interests that overlap with your secondary interests

Example of Code Bidding Rochester, MN Marshalltown, IA Waterloo, IA 283 E 378 D 452 E Round McLeod USWest McLeod USWest AT&T McLeod USWest 24 56,000 287,000 … … … 46 568,000 52 689,000 55 723,000 58 795,000 59 875,000 313,378 60 345,000 62 963,000 64 62,378 1,059,000 65 69,000 68 371,000

Critical times

Critical times • Auction opening • Stage transitions • Auction closing

decision making Bidder

Bidder decision making • Valuation model • Competitor analysis • Bid tracking tool • Residual supply • Communication with investors • Deviations from rational bidding

Valuation model • Typically complex spreadsheet, calculating cost and revenues over 10 year horizon (NPV) • Most sophisticated models reconfigure network in response to extra spectrum

Bid tracking tool

Bid tracking tool: Access and Excel • Import round (download results) • Market analysis • Bidder analysis • Bid form (upload bids) • Map • Rounds summary • Market pivot table • Bidder pivot table

Italy 3G STRATEGY ROUND 11 Basic/Enhanced relates to E2 Blu = lowest {Blu, Ipse, Andala} E1 Andala = 2nd lowest {Blu, Ipse, Andala} Minumum Bid 492 JUMP WARNINGS INACTIVITY WARNINGS NEW ENTRANTS & BLU COMPARISON OPI position round bidder jump Highest. Above Ipse, Andala and Blu. TIM for 2 rounds BLU not moved above Andala for 1 round Rationale (Enhanced default strategy) BLU for 3 rounds BLU not moved above Ipse for 2 rounds No bid. Case 11: Do not have to bid, above E1. Epsilon 0 Basic bid + epsilon No bid Net value of extra bid N/A Extra bid 0 Enhanced bid No bid Enhanced bid + epsilon No bid Pr(Andala quits before Ipse) 0.50 Pr(Blu quits before Ipse) 0.50 Pr(Blu quits before Andala) 0.50 BLU+1 END GAME STRATEGY (suggestion) Suggest high pr(Blu quits before Ipse) and high pr(Blu quits before Andala) Suggested bid No bid BLU failed to move above Andala for 1 round. BLU failed to move above Ipse for 2 rounds. BLU inactive for 3 rounds. Round results Ranking Minimum R Must bid Tim Omnitel Blu Wind Ipse Andala 1 2 3 4 5 6 bid 0 400 400 400 400 400 400 400 Ipse 400 Omnitel 400 Tim 400 Blu 400 Wind 400 Andala 400 1 420 Andala 400 400 400 420 400 420 Andala 420 Wind 420 Ipse 400 Omnitel 400 Tim 400 Blu 400 2 420 Blu 400 423 421 422 400 420 Omnitel 423 Wind 422 Blu 421 Andala 420 Ipse 400 Tim 400 3 420 Tim 422 425 421 422 422 420 Omnitel 425 Wind 422 Tim 422 Ipse 422 Blu 421 Andala 420 4 443 Andala 422 425 421 422 422 443 Andala 443 Omnitel 425 Wind 422 Tim 422 Ipse 422 Blu 421 5 443 Blu 422 425 444 422 422 443 Blu 444 Andala 443 Omnitel 425 Wind 422 Tim 422 Ipse 422 6 444 Ipse 422 425 444 422 445 443 Ipse 445 Blu 444 Andala 443 Omnitel 425 Wind 422 Tim 422 7 444 Tim 444 448 449 447 445 446 Blu 449 Omnitel 448 Wind 447 Andala 446 Ipse 445 Tim 444 8 468 Tim 468 448 449 447 468 446 Tim 468 Ipse 468 Blu 449 Omnitel 448 Wind 447 Andala 446 9 470 Andala 468 448 449 470 468 470 Wind 470 Andala 470 Tim 468 Ipse 468 Blu 449 Omnitel 448 10 472 Omnitel 468 474 449 470 473 470 Omnitel 474 Ipse 473 Wind 470 Andala 470 Tim 468 Blu 449 11 492 Blu