Game Theory

for
Strategic Advantage

15.025

Alessandro Bonatti
MIT Sloan
Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 1
 Part III: “Big” Applications

Repeated Asymmetric Communication

Interaction Information

Long-Run Designing Credibility &

Relationships Auctions & Markets Reputation

Classes 11-14 Classes 15-18 Classes 19-21

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 2

 Games Played over Time
Today:

• Price wars

• Dynamic Pricing

• Loyalty Programs

Beginning next class:

• Repeated games

• Theory and cases

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 3

 NYC Tabloids

• July 1994: NY Post “tested” a price of 25 cents

• Daily News price = 40 cents. (Yes, it matters!)

• Why? NY Post previously went from 40 to 50.

• Daily News “did not cooperate” (= follow).

• NY Post cut price $ 0.25 in Staten Island only.

• Shortly afterwards, Daily News priced at 50 cents!

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 4

 Airline Pricing

• History of anticompetitive “devices”

• Reservations system

• Frequent-flier programs (AA, 1981)

• Miles? Points?? Dollars???

• Hub-and-spoke model
• Direct flights vs. connecting flights prices

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 5

 Dynamic Pricing Game
Long game, tricky logic: think through every branch

1. Separate role for signals and loyalty programs

2. Keeping the game manageable ( your projects!)

3. Backward Induction refresher

Read  Play online  Solve  Discuss  Extend

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 6

 Dynamic Pricing: Rules

• Two firms. 100 customers. Zero marginal costs.

• Stage 1: Invest in creating “Loyal” customers?
– Loyal customers buy from you no matter the price
– Two choices: 0 Loyal (no cost) or 30 Loyal (cost $250)

• Stage 2: Firms alternate price announcements.

– May only announce “cuts” or “confirm previous price”
– Choices are $50, $40, $30, $20, $10

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 7

 Dynamic Pricing: Rules

• Sales are made only after prices “settle”

• Potentially very long game
• Can be played with several goals in mind

• Simplified Format
• Game ends when a player confirms previous price
• Play to maximize monetary payoffs

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 8

 Dynamic Pricing: Payoffs
• Payoffs = Revenue – Loyalty Cost
• If prices settle at $50 for both firms, the “disloyal”
customers are split evenly.
• Otherwise, firms do not split the disloyals equally:
– “Larger firm” is one that (a) has lower price, or
(b) was first to announce final price (if equal)
– Larger firm sells 100 or 70 at its own price
(depending on loyalty of other firm’s customers)
– Smaller firm sells 0 or 30 at its own price

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 9

 Preparation Questions
• You have invested in Loyal customers, but the other firm
has not. They win the coin flip and choose price $40.
– How would you respond? Why?

• Neither you nor your opponent has invested in Loyal

customers. You win the coin flip.
– Do you begin at $50 or go lower? How much lower?

• Will you invest in Loyal customers or not?

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 10

 Game-Theoretic Analysis

• Construct solution taking the game “as-is”

• Reason through various scenarios

• Conclusions for investment in Loyalty

• Modify the game / discuss alternative strategies

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 11

  Look Forward, Think Back 
• Two-stage games: what happens in the 2nd stage?

• Evaluate ALL possible scenarios, ROLL BACK

• Build a “Game Outlook” from stage 1

• Key principle: backwards induction

Prof. Alessandro Bonatti 12 MIT Sloan 15.025 Spring 2015

 Backwards Induction
• Loyalty strategies are public after Stage 1

• In Stage 2, a different game for any combination of

“Loyal” and “Not Loyal” … (L,NL), (L,L), (NL,NL), (NL,L)

• Compute revenues in all scenarios

• Game Tree?

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 13

 Multi-Stage Game
Firm 2
STAGE 1
L NL
L ?-250 , ?-250 ?-250 , ?
Firm 1
NL ? , ?-250 ? , ?

STAGE 2
Dyn Pricing: Dyn Pricing: Dyn Pricing: Dyn Pricing:
(1) Loyal vs. (1) Not Loyal vs. (1) Not Loyal vs. (1) Loyal vs.
(2) Loyal (2) Loyal (2) Not Loyal (2) Not Loyal

Compute all Stage 2 Revenues and roll back!!!

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 14

 Upside of Loyalty
• If other firm undercuts you, your final payoff will
be higher if you secured some loyal customers

• Suppose the other firm has no loyal customers

and begins at $40…

• What would you do if you do NOT have loyalty?

• And what if you DO?

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 15

 Disloyal vs. Disloyal  $40
$50 Game ends. Your Rev = $0

$40
Since each firm has no loyal customers,
$30 each will re-undercut until price equals $10.
If you bid $30 or $20, other will go to $10
$20 and you will get zero revenue

Game ends. Your Rev = $1000

$10

Any disloyal must be first to the lowest price

Respond with $10 and get $1000
Not as good as $1500 if you had Loyal
Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 16
 Lesson 0

Look all the way ahead in the game

(Don’t respond to $40 with $30!)

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 17

 “Loyal” vs. “Disloyal” $40
Pricing stage (you are “Loyal”)  the $250 is sunk
Game ends. Your Rev = $1500
$50
Other firm stays at $40.
$40 Your Rev=$40*30=$1200
If other firm responds with $20, you undercut
$30 to $10 (you prefer 100*$10 to 30*$30.)
So, other firm goes right to $10, ending the
$20 game and your rev = $30*30 = $900
Other will undercut to $10.
Your Rev = $20*30 = $600
$10 Game ends. Your Rev = $1000

• Disloyal must be the low-price firm (or 1st to a tie)!

• Look at each step and conclude that $50 is best!
Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 18
 Downside of Loyalty
• A “disloyal” opponent can undercut you without
triggering a price war!
• Unwillingness to re-undercut makes you an easy target
– Disloyal opponent (whether first or second) will
undercut you with $40, leaving you with only $1500
• A “Loyal” opponent (if they go first) also undercuts you
with $40… why?

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 19

 “Loyal” Undercut by “Loyal”$40

Game ends. Your Rev = $1500

$50
Other firm stays at $40.
$40 Your Rev=$40*30=$1200
Other firm responds with $20, you prefer
$30 30*$30 to 70*$10. Other firm goes to $20,
$20 your rev = $30*30 = $900

Other will stay at $40.

Your Rev = $20*70 = $1400
$10 Game ends. Your Rev = $700

• Look at Each Step and conclude that $50 is best!

• But then 1st-moving Loyal gets Rev = $40*70 = 2800!
Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 20
 Upside of Disloyalty

• You are so “Lean & Hungry” that no Loyal opponent

messes with you

• Against Loyal opponent, you go to $40

(no matter who goes first), get Rev = 70*$40 = $2800

• What about against Disloyal?

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 21

 Disloyal vs. Disloyal
• Any undercutting leads to ultimate price of $10
• Revenues are no greater than $1000
• Better to stick with $50 VERY FRAGILE !!
• No price war in equilibrium!
$50
$50 2500 , 2500

$10
0 , 1000
$40 $50
4000 , 0
$40
4000 , 0
$10
$10 0 , 1000
1000 , 0
Lean and Hungry works well!
Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 22
 Creating Loyalty: Two Effects
• Direct effect of Loyalty: secure part of the demand
– Direct effect stronger if program is more attractive
– Stronger if you can price discriminate?

• Strategic effect of Loyalty: weaker bargaining position

– Doesn’t matter in a price war
– Matters more if undercutting is “on the margin”

• Which one is stronger?

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 23

 Multi-Stage Game
Firm 2
STAGE 1
L NL
L ?-250 , ?-250 1500-250,2800
Firm 1
NL 2800,1500-250 2500,2500

STAGE 2
Dyn Pricing: Dyn Pricing: Dyn Pricing: Dyn Pricing:
(1) Loyal vs. (1) Not Loyal vs. (1) Not Loyal vs. (1) Loyal vs.
(2) Loyal (2) Loyal (2) Not Loyal (2) Not Loyal

Loyal vs. Loyal: first mover cuts to $40

Expected revenue = 0.5*2800+0.5*1500 = 2150
Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 24
 Reduced-Form Game
Firm 2
STAGE 1
L NL
L 1900 , 1900 1250 , 2800
Firm 1
NL 2800 , 1250 2500 , 2500

Not Loyal is a Dominant Strategy!

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 25

 Lesson 1

Keep track of the

Direct and Strategic Effects!

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 26

 Retail Competition
• Nordstrom competing in the cutthroat world of online retail

• Nordstrom price-match guarantee:

“We are unable to match prices from auction and outlet stores
or their websites, or other retailers’ discount promotions,
shipping offers and gift card offers.”

• Focused on retailers perceived as direct competitors.

• Commitment to aggressive behavior (the opposite of Loyalty!)

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 27

 Creating Loyalty: a Bad Idea?
• Loyalty is a dominated strategy in stage 1!

• Why would you go ahead anyway?

– Fear of price-wars (fragile equilibrium)

– Fraction of “captive” customers

– Price discrimination

– Collusion on prices

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 28

 Fear of Price War
• If you fear p = $10 in Not Loyal vs. Not Loyal “subgame” …
• … coin flip determines who makes all the sales
• Expected payoffs in stage 1 look like this

Firm 2
L NL
L 1900 , 1900 1250 , 2800
Firm 1
NL 2800 , 1250 500 , 500

• Two Nash Equilibria – as in the cities game

• “Free-riding on insurance against price war”
• Stable population split = 45% Loyal

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 29

 Lesson 2

Keep in mind your opponents’ goals

(and / or doubt their rationality)

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 30

 Ryanair
• Launched Dublin-London route in 1986

• Enter a market with similar choices as incumbents?

– Aer Lingus “to bring benefits to the Irish community”
– British Airways has (obviously) deep pockets.
• Lowest BA fare was GBP 99
• Ryanair enters at GBP 98, “first-rate customer service
and on-board service comparable to BA.”

• What happened? How did the game change?

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 31
 Captive Consumers
• If 50 customers can be “captured” then strong direct effect
• Loyal vs. Disloyal still ends $50:$40, but different market shares
• Expected payoffs in stage 1 look like this

Firm 2
L NL
L 2250,2250 2250,2000
Firm 1
NL 2000,2250 2500,2500

• Two Nash Equilibria – as in the “Stag Hunt”

• If I think you’re building loyalty, I must protect my market
• UnStable population split = 50% Loyal

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 32

 Lesson 3

Relative strength of direct vs. strategic

effects shapes the game
(and potentially the industry)

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 33

 Price Discrimination
• Back to 30 captive customers
• If you could charge different prices to frequent and occasional
travelers… Expected payoffs in stage 1 would look like this
Firm 2
L NL
L 2250,2250 3000,1750
Firm 1
NL 1750,3000 2500,2500

• Price wars could occur in every subgame

• All prices = $50!! Pure business-stealing! Prisoners’ Dilemma!!!!!!
• What actually happened?
• [Same outcome if prices were “fixed” – collusion]
Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 34
 Lesson 4

Think the whole game through

the (credible) threat of (price) war keeps all

prices high…

… absent price discrimination or

promotions, loyalty programs may
undermine the ability to threaten…

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 35

 Airline Loyalty Programs
Co-opetition, Ch.5 (posted)

• American starts AAdvantage

• United copies MileagePlus

• Most others follow

• Low-cost carriers typically more hesitant

• Price discrimination? Like GM and Ford credit cards…

• Tacit(?) coordination / reservation systems / price fix

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 36

 Takeaways
• Look forward, think back
• Subgame-perfect equilibrium key tool for dynamic games
• Rationality of opponent (undercutting machine?)
• Direct vs. strategic effect:
– Protect demand
– Weaker bargainer
• Relative strength of the 2 effects (# of captives)
• Credible threats (could cut 1 price only, but..)

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 37

 Next Time

• Kick off from today’s game 

 Repeated interaction

• Read: “The Dynamics of Price Competition”

(it’s about tit-for-tat, etc…)

Prof. Alessandro Bonatti MIT Sloan 15.025 Spring 2015 38

MIT OpenCourseWare
http://ocw.mit.edu

15.025 Game Theory for Strategic Advantage

Spring 2015

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.