lecture 1112

39
Game Theory “A Cynic Knows the Price of Everything and the Value of Nothing” -Oscar Wilde, Lady Windemere’s Fan Mike Shor Lectures 11 & 12

Upload: gd3000

Post on 25-Dec-2015

258 views

Category:

Documents


0 download

DESCRIPTION

teoria de juegos

TRANSCRIPT

Page 1: Lecture 1112

Game Theory

“A Cynic Knows the Price of Everything and the Value of Nothing”

-Oscar Wilde, Lady Windemere’s Fan

Mike ShorLectures 11 & 12

Page 2: Lecture 1112

Game Theory - Mike Shor 2

What is an Auction?

auc•tion1. A public sale in which property or

merchandise are sold to the highest bidder.

2. A market institution with explicit rulesdetermining resource allocation and prices on the basis of bids from participants.

3. Games: The bidding in bridge

[Latin: auctiō, auctiōn- from auctus, past participle of augēre, to increase]

e

(ôk’sh n)

Page 3: Lecture 1112

Game Theory - Mike Shor 3

Examples of Auctions FCC Spectrum Procurement Auctions Treasury Bills Internet Wine Options Quota Rights

Page 4: Lecture 1112

Game Theory - Mike Shor 4

Classifying Auctions Private value auction

• Each bidder’s value is known to him or her• Values of other bidders unknown

Common value auction• Item has single, true value• Value unknown at time of bidding

Page 5: Lecture 1112

Game Theory - Mike Shor 5

Private Value Dinner.

Page 6: Lecture 1112

Game Theory - Mike Shor 6

Common Value Unproven oil fields.

Page 7: Lecture 1112

Game Theory - Mike Shor 7

Basic Auction Types Open Auctions (sequential)

• English Auctions• Dutch Auctions• Japanese Auctions

Sealed Auctions (simultaneous)• First Price Sealed Bid• Second Price Sealed Bid

Page 8: Lecture 1112

Game Theory - Mike Shor 8

English Auction Ascending Bid

Bidders call out prices (outcry) Auctioneer calls out prices (silent) Bidders hold down button (Japanese)

Highest bidder gets the object Pays a bit over the next highest bid

Page 9: Lecture 1112

Game Theory - Mike Shor 9

Dutch (Tulip) Auction Descending Bid “Price Clock” ticks down the price

First bidder to “buzz in” and stop the clock is the winner

Pays price on clock

Page 10: Lecture 1112

Game Theory - Mike Shor 10

Sample Dutch AuctionMinimum Bid: $10

$120$110$100$ 90$ 80$ 70

$ 10$ 20$ 30$ 40$ 50$ 60

Page 11: Lecture 1112

Game Theory - Mike Shor 11

Sealed Bid Bidders submit secret bids to the

auctioneer The auctioneer determines the

highest and second highest bids

Highest bidder gets the object and• Pays own bid (First-Price)• Pays the next highest bid

(Second-Price Vickrey)

Page 12: Lecture 1112

Game Theory - Mike Shor 12

Strategic Equivalence of Auction Types

Dutch Auction~ First Price Sealed Bid

English Auction ~ Second Price Sealed Bid

Page 13: Lecture 1112

Game Theory - Mike Shor 13

Other Auction Formats Double auction

• Buyers and sellers bid• Stock exchanges

Reverse auction• Single buyer and multiple sellers• Priceline.com

Multiunit auction• Seller has multiple items for sale• FCC spectrum auctions

Page 14: Lecture 1112

Game Theory - Mike Shor 14

The Vickrey Second Price Auction Bidding strategy is easy!

Bidding one’s true valuation is a (weakly) dominant strategy

• Proxy bidding on eBay

Intuition: the amount a bidder pays is not dependent on her bid

Page 15: Lecture 1112

Game Theory - Mike Shor 15

Optimal Bidding StrategySecond Price Auctions

You Lose

You Win

hig

her

Your bid

Others’ bids

Your value

Page 16: Lecture 1112

Game Theory - Mike Shor 16

Bidding Higher Than My Valuation

Case 1 Case 2 Case 3

No difference No difference Lose money

Page 17: Lecture 1112

Game Theory - Mike Shor 17

Bidding Lower Than My Valuation

Case 1 Case 2 Case 3

No difference No difference Lose money

Page 18: Lecture 1112

Game Theory - Mike Shor 18

First Price Auction Second price auction

• Always bid your true valuation

First price auction presents tradeoffs• If bidding your valuation – no surplus• Lower your bid below your valuation

Smaller chance of winning, lower price• Bid shading

Depends on the number of biddersDepends on your informationOptimal bidding strategy complicated!

Page 19: Lecture 1112

Game Theory - Mike Shor 19

Revenue Equivalence Assume an auction with the

following rules:• The prize always goes to the person with

the highest valuation• A bidder with the lowest possible valuation

expects zero surplus

All such auctions yield the same expected revenue!

Page 20: Lecture 1112

Game Theory - Mike Shor 20

Revenue in the Real World Risk Aversion

• Does not influence 2nd price auctions• Risk averse bidders are more aggressive in

first-price auctions

Non-familiarity with auctions• More overbidding in second-price auctions• More overbidding in sealed-bid auctions

Page 21: Lecture 1112

Game Theory - Mike Shor 21

More Bidders More bidders leads to higher prices More bidders leads to less surplus

Example (second price auction): Each bidder has a valuation of either $20 or $40, each with equal probability

Page 22: Lecture 1112

Game Theory - Mike Shor 22

Number of Bidders Two biddersPr{20,20}=Pr{20,40}=Pr{40,20}=Pr{40,40}= ¼

Expected price = ¾ (20)+ ¼ (40) = 25

Three biddersPr{20,20,20}=Pr{20,20,40}=Pr{20,40,20}

= Pr{20,40,40}=Pr{40,20,20}=Pr{40,20,40}= Pr{40,40,20}=Pr{40,40,40}= 1/8

Expected price = (4/8) (20) + (4/8) (40) = 30

Page 23: Lecture 1112

Game Theory - Mike Shor 23

Number of Bidders Assume more generally that valuations

are drawn uniformly from [20,40] Expected price:

20

25

30

35

40

1 10 100 1000

Page 24: Lecture 1112

Game Theory - Mike Shor 24

Designing Auction Rules Bidding

• Who is allowed to bid?• How are bids presented?• How much must bids be beaten by?• Is bidding anonymous or favored?• Is there a reserve price?

Information• Are current bids revealed?• Are winners identified?

Clearing• Who gets what?

Page 25: Lecture 1112

Game Theory - Mike Shor 25

“Tweaking the Rules” IeBay …

Three laptops for sale Top three bidders pay

the third highest bid Opening bid: $1 Current high bids:

$50, $80, $400 How high should next bid be?

Page 26: Lecture 1112

Game Theory - Mike Shor 26

“Tweaking the Rules” IIFCC Spectrum Auctions… Want to encourage minority and

female-owned firms to bid but licenses are very expensive.

Reserve “blocks C & F” for smaller bidders.

Allow 10% down, low interest, remaining principal owed in 7 years.

What happens?

Page 27: Lecture 1112

Game Theory - Mike Shor 27

FCC Spectrum Auctions Bid high – if licenses end up being worth

less, default! 83 winners: Of the four largest, … … went bankrupt and defaulted … got $1B reduced to $66M in

bankruptcy court … was a front for Qualcom … was sold to Siemens

Page 28: Lecture 1112

Game Theory - Mike Shor 28

“Tweaking the Rules” IIIFCC Spectrum Auctions… Largest potential problem: Collusion

• Bid rotation (construction)• Subcontracting (Treasury Bills)

Discouraging Collusion• Sealed-bid Auctions• Anonymity – do not release results

“I bid $16,000,483” “I bid $11,000,212”

Page 29: Lecture 1112

Game Theory - Mike Shor 29

Sources of Uncertainty Private value auction

• Inherent differences among bidders• Item is for personal use – not resale• Values of other bidders unknown

Common value auction• Item has single, true value• Value unknown at time of bidding

Page 30: Lecture 1112

Game Theory - Mike Shor 30

Common Value Auctions Example: Offshore oil leases

• Value of oil is same for every participant• No bidder knows value for sure• Each bidder has some information• Exploratory drilling

Different auction formats are not equivalent

• Oral auctions provide information• Sealed-bid auctions do not

Page 31: Lecture 1112

Game Theory - Mike Shor 31

Levels of Thinking

What would I be willing to pay given

what I know before submitting my bid

versus

what I know before submitting my bid, and

that I will only win if no one else is willing to bid higher than me

Page 32: Lecture 1112

Game Theory - Mike Shor 32

Common Value Auction Learn about value from other bids

Estimates on average correct

Winner not picked randomly – highest estimate -> too high

Page 33: Lecture 1112

Game Theory - Mike Shor 33

Avoiding Winner’s Curse

COMMANDMENT

The expected value of the object is irrelevant.

To bid: Consider only the value of the object if you win!

Page 34: Lecture 1112

Game Theory - Mike Shor 34

Avoiding Winner’s Curse Bidding with no regrets:

Since winning means you have the highest signal, always bid as if you have the highest signal

If you have highest signal – what is the object worth?

Use that as the basis of your bid

Page 35: Lecture 1112

Game Theory - Mike Shor 35

Private or Common Value? Most Auctions are not purely private

value or purely common value

Private Value• Resale component (art)• Reputation / conspicuous consumption

Public Value• Differing abilities (mining)

Page 36: Lecture 1112

Game Theory - Mike Shor 36

Multiple Unit Auctions If products are distinct and

valuations are independent, can hold distinct auctions

If values are not independent…• Objects interrelated• Substitutes• Complements

… then things get very messy!

Page 37: Lecture 1112

Game Theory - Mike Shor 37

Dependent Multi-Unit (Combinatorial)

PCS Spectrum Auctions

Page 38: Lecture 1112

Game Theory - Mike Shor 38

3x108 m 0 HZ

3x10-7 Å 1025 HZ

Very Low Frequency

(VLF)

LF HF UHF EHF MF VHF SHF

Infrared Visible Ultraviolet X-Ray Gamma Ray Cosmic Ray

3 KHZ 300 GHZ

Marine AM Radio Amateur VHF UHF Misc Satellite

UNLICENSED B E F C D A B E F C D A

1850 MHZ 1910 1930 1990 MHZ

300 MHZ 3 GHZ

TV Broadcasting Aeronautical Channels 14-36 Channels 38-69 Radionavigation Misc Misc

18 winning bidders, 99 licenses

$7.7 billion dollars

Page 39: Lecture 1112

Game Theory - Mike Shor 39

Summary Participating

• Bid true valuation in 2nd price auctions• Shade bids in 1st price auctions• In common value auctions, assume that you

have the most optimistic estimate Designing

• Preclude cooperation among bidders• Announced reserve price to gain efficiency

and relieve doubts• In common value auctions, provide

information to help with winner’s curse