Upload
gd3000
View
214
Download
0
Embed Size (px)
DESCRIPTION
teoria de juegos
Citation preview
Game Theory
The Power to Constrain an Adversary Depends Upon the Power to Bind Oneself.- Thomas Schelling
Mike ShorLectures 7&8
Game Theory - Mike Shor
ReviewCooperation requires sacrificing immediate profits for a future relationship
The sacrifice is only made if the punishment is severe enough
Punishment that is too severe is not credible
Game Theory - Mike Shor
Finite RepetitionUnraveling prevents cooperation if the number of periods is fixed and knownProbabilistic terminationThe game continues another round with some probability p:Equivalent to infinite game$1 next period is worth p todayValue of future ={ value if there is a future }
{ probability of a future } Effective interest rate: r = (1+r)/p 1
Game Theory - Mike Shor
LessonsBe careful of finite games
Always a chance of another encounter
Balance severity of punishment!
Mild punishment may not deter cheatingSevere punishments may not be credible
Game Theory - Mike Shor
Credibility & Commitment The difference between genius and stupidity is that genius has its limits.Albert Einstein
The difference between 'involvement' and 'commitment' is like an eggs-and-ham breakfast: the chicken was 'involved' - the pig 'committed'.Unknown
Game Theory - Mike Shor
Talk is CheapPromises
Continental Airlines said yesterday that it would raise airfares on about two-thirds of its routes to take effect September 5. - The New York Times August 29, 1992
Continental Airlines has dropped its plan to raise domestic airfares by 5%. - USA Today September 4, 1992
Game Theory - Mike Shor
Talk is Cheap Threats
On January 5, Boeing, the worlds top aircraft maker, announced it was building a plane with 600 to 800 seats, the biggest and most expensive airliner ever. Some in the industry suggest Boeings move is a bluff to preempt Airbus from going ahead with a similar plane. - Business Week, 1993
Game Theory - Mike Shor
And Getting Cheaper Airbus announces commercial launch of the A3XX, the largest civil aircraft ever built.
Boeing has said that there is no market for such a large plane and has decided to modernize its trustworthy 747 family of planes rather than build its own megaseater. - Associated Press June 23, 2000
Game Theory - Mike Shor
Conventional Wisdom
Dont burn bridges.Decrease downside risk.It is nice to have more options. While you may have money to burn, you shouldnt burn money
Game Theory - Mike Shor
Strategic CommitmentYou are not credible if you propose to take suboptimal actions.
Specifically, if a rational actor proposes to play a strategy which earns less than maximal profit.
How to be credible?
Game Theory - Mike Shor
CredibilityRemove strategiesfrom your own set of future choices the strategies that may tempt you in the futureGiving away your patent?Reduce payoffs from those strategies that may tempt youCustomers as hostages.Become irrationalRemove human meddling
Game Theory - Mike Shor
A Non-credible ThreatTen suppliers each have two options:Deliver on time at a cost of $70,000Deliver a week late at a cost of $20,000Delivery results in $100,000 payment
I need at least nine suppliers and the suppliers know this
Game Theory - Mike Shor
A Non-credible ThreatI threaten not to deal with a late supplier (at most one)
Two equilibria:All deliver on timeAll deliver late (the likely equilibrium)Even if only one other supplier delivers late, it is in my best interest to do so: (80) > 30
Game Theory - Mike Shor
A Credible ThreatI number suppliers (110) and refuse delivery from the lowest numbered one among those who are late.Supplier 1 delivers on time (better than getting nothing)Thus, supplier 2 delivers on time
Game Theory - Mike Shor
Removing Strategies IDelegation
In contract negotiation, can squabble over many detailsInstead, send an agent with power of attorney to sign as is or walk awayHaggling over prices in a department store
Learn from government bureaucracy: The rules wont allow me to do what you ask
Game Theory - Mike Shor
Delegation ExamplesHuman resource departmentsShield from requests for higher salariesHR execs not compensated based on employer valueCollection agenciesshield from pleas or threats of the debtorreinforce repayment to protect reputationAccounting firms overseeing contestsAccountants payment not tied to outcomeConcerned with reputation for fairness
Game Theory - Mike Shor
Removing Strategies IIBurning BridgesPower comes from not being able to retreatAllow opponent to retreat (Sun Tzu)
The PatchHunt for Red OctoberCortes upon arriving in Mexico
Game Theory - Mike Shor
Burning Bridges(Example 1)Semiconductor patent sharing
Mosaid Technologies, a designer and licensor of semiconductor chips and technologies, just announced a patent sharing deal with Mitsubishi Electric
Share patent with another competing firmCommit to chip supply to production plantsCommit to no opportunistic behavior
Game Theory - Mike Shor
Burning Bridges(Example 2)Capacity constraints
New entrant commits to low productionThe puppy-dog ploy
Puppy Dog PloyStay weak to avoid an aggressive response
Apple v. Microsoft
Game Theory - Mike Shor
Removing Strategies IIIIrrationality
U.S. / U.S.S.R. nuclear deterrence Mutually Assured Destruction (MAD)like Grim Trigger Strategy Proportional Responselike Tit-for-TatWant a lot of deterrenceWant irrationality to be credibleDr. Strangelove & the Doomsday device
Game Theory - Mike Shor
Dr. StrangeloveSeverity
Create fear in the mind of the enemyIrreversibility
Must be irreversibleIrrationality
Not something a sane man would doPracticality
Punishment shouldnt be too harshClarity
Tell the world
Game Theory - Mike Shor
Use of ContractsPromises vs. Threats
Promises can sometimes be credible through a contract with the party to whom you are making the promiseNot always: Mario Puzos The GodfatherThreats can never become credible by use of a contract with the party you are threateningMust contract with third party
Game Theory - Mike Shor
Reducing PayoffsContracting with customers to commit to competitors
Price-matchingMost Favored Customer clausesContracting with lenders to commit to a take-over
Interest-rate rise if loan amount increases
Game Theory - Mike Shor
Example From Long Ago Two firms: Firm 1 and Firm 2Two prices: low ($4) or high ($5 )3000 captive consumers per firm4000 floating go to firm with lowest price
Firm 2LowHighFirm 1Low 20 , 20 28 , 15High 15 , 28 25 , 25
Game Theory - Mike Shor
Contracting with CustomersThe game is a prisoners dilemmaBoth firms prefer:{High,High}Only equilibrium: {Low , Low}Cannot credibly promise to play HighEven if committed to High, other firm would still respond with LowHow to resolve this?Third party contracts with customers
Game Theory - Mike Shor
Most Favored CustomerSay in period 1, the firms colluded and each sold to 5000 customers
In period 2, firms must refund to last periods customers $1 each if price is low
What is the impact on the game?
Game Theory - Mike Shor
Most Favored Customer
Firm 2LowHighFirm 1Low 20 , 20 28 , 15High 15 , 28 25 , 25
Firm 2LowHighFirm 1Low 15 , 15 23 , 15High 15 , 23 25 , 25
Game Theory - Mike Shor
Contracting with LendersTakeover offer: $200 millionYou can afford $20 million / yearFinance takeover for 20 years at 7%Add penalty: if amount greater than $200 million, +1.5 points on interest rateAnnual Payments:
$200 million:$18.6 million / year$210 million:$19.6 million / yearwith penalty: $21.9 million / year
Game Theory - Mike Shor
Extravagance in Nature
Game Theory - Mike Shor
Extravagance in Business
Game Theory - Mike Shor
IrrationalityBurning money
Commit to long-term market planCommit to high quality productsFirm can produce high or low qualityCustomers cannot observe quality
Once customers buy a product the first time, observe quality, punish dishonest firms
QualityMarginal costValueHigh$8$50Low$4$20
Game Theory - Mike Shor
Burning MoneyProducing high quality is more profitable
$50 - $8 = $42 > $16 = $20-$4Pretending you are high quality is best
$50 - $4 = $46Incentive to lie:
Customers do not believe in high qualityCustomers only pay for low qualityFirm only produces low quality INEFFICIENT AND UNPROFITABLE
Game Theory - Mike Shor
Burning MoneyAnnounce high quality and burn $50
QualityProfitMoney BurnedNet ProfitHigh$42/year$50$8 loss this year, $42 in the futureLow$46 once$50$4 loss this yearNo Entry$0$0$0
Game Theory - Mike Shor
Burning MoneyBy burning $50, convince customers of your high quality:
$8 loss this year, $42 profit in the futureWithout burning money, can only sell low quality:
$16 profit this year and in the futureGaining credibility by burning $50
Invest in future relationship with customers
Game Theory - Mike Shor
Forcing Your OpponentSimilar tactics for making your opponent commit to strategies in your favorIncreasing his strategy space
Excluding bargaining agentsLowering his payoffs
Poison pillsRaising his payoffs
Reputation bolstering
Game Theory - Mike Shor
Price MatchingIf one firm charges low, it does not gain any additional customers, since the competitor automatically matches it.
What is the effect on the game?
NASDAQ order preferencing?
Game Theory - Mike Shor
Price Matching
Firm 2LowHighFirm 1Low 20 , 20 28 , 15High 15 , 28 25 , 25
Firm 2LowHighFirm 1Low 20 , 20 20 , 20High 20 , 20 25 , 25
Game Theory - Mike Shor
Commitment Is Counterintuitive COMMANDMENT.
Reduce your strategy space and decrease your own payoffs to commit.
Increase your opponents strategy space and alter your opponents payoffs to preclude the rival from committing.Hurt yourself to help yourselfHelp your opponent to help yourself
Game Theory - Mike Shor
CommitmentUnder UncertaintyAn offer you cant refuseAfter a seemingly successful interview, the interviewer asked where the firm ranks on your list of potential employeesBefore answering, you are told:
The firm only hires applicants who rank it firstIf the firm is in fact your first choice, then you must accept a job offer in advance, should one be made
Game Theory - Mike Shor
Commitment Under UncertaintyWhat to do?
Binding early-decision college applications
Why make such proposals?
Take advantage of your uncertaintyTake advantage of your risk-aversionMake you commit before they do!
Game Theory - Mike Shor
Flexibility vs. CommitmentFlexibility in light of uncertainty about eventual outcomes generates value
Keeping your options openMust be balanced against strategic value of commitmentOption Value:
The additional expected profit from remaining flexible above the expected profit earned from committing
Game Theory - Mike Shor
ExampleOption Value of DelayA firm can spend $100 million on an investment to enter a new marketMarket demand uncertain:
High acceptance: revenues of $300m(Probability=0.5)Low acceptance: revenues of $50m(Probability=0.5)Two options:Invest today in the presence of uncertaintyWait a year for full revelation of information
Game Theory - Mike Shor
Calculating Option Value Immediate investment:
E[]=(1/2)(300-100) + (1/2)(50-100)= $75 million Delayed investment:
Only invest if high acceptanceE[]=(1/2)(300-100) + (1/2)(0) = $100 million Option value:
$100 - $75 = $25 million
Countervailing force: by waiting, the firm risks having the opportunity pre-empted by competitors
Game Theory - Mike Shor
Philips, N.V.Capacity commitment in CD introductionPhilips: innovators advantage
Initiate construction of plant ahead of competitorsDecision problem of Philips in 1982:
Build a disk-pressing plant in the U.S. And invest in a substantial amount of capacity to deter potentially entry (Sony, etc.)Delay decision until commercial appeal of CDs can be determined. Import CDs to the U.S. To test the waters.
Game Theory - Mike Shor
Option Value: Three Casesq probability of mass acceptance of CDs
Monopoly Benchmark:
Philips should wait if q < 0.380Sony competition, equal information
Philips should wait if q < 0.006Sony competition, better information
Philips should wait if q < 0.130
Game Theory - Mike Shor
SummaryPure option value
In the absence of competition, Philips would have been better off waiting and retaining flexibility if the probability of acceptance was 0.38 or lowerCommitment value
Faced with competitors who are as well informed, Philips would be better off building the U.S. plant right away even in the presence of uncertaintyInformational advantage
Given proprietary information through its CD operations in Europe, Philips should remain flexible if the probability of acceptance was 0.13 or lower
Game Theory - Mike Shor
EvidencePhilips did not build a U.S. Plant in 1983
Its assessment of the likelihood of general acceptance did not meet the thresholdMarket realization (surprise!)Sony constructed a U.S. Plant in 1984
Terry Haute, IndianaPhilips attempted to compete
Increased capacity in Hanover, Germany plantPhilips decided to invest in a U.S. Plant
Only after the Sony plant was fully operational
Game Theory - Mike Shor