Lecture Notes 10

Game Theory I

Xu Le
National University of Singapore
I. Sequential move Games
Bertrand Paradox
Ø  The previous example shows that firms are making zero profit: Even a
duopoly would suffice to restore the welfare-optimal market outcome as
under perfect competition.
 
─  Hard to believe that two firms are enough to entirely eliminate market
power – thus called “Bertrand Paradox”.

Ø  The paradox can be solved by relaxing some of the assumptions made for the
model. Among others,
 
─  Asymmetric costs
─  Capacity constraints
─  Product differentiation (Refer to the next slides)
─  Repeated interaction
─  Incomplete information
Price Competition with Differentiated
Products
Suppose each of two duopolists has fixed costs of $20 but zero variable costs,
And they face the following demand curves:
Q1 =12 − 2P1 + P2
Q2 =12 − 2P2 + P1
Choosing prices

Firm 1’s profit: π = P ×Q −TC

1 1 1

Firm 1’s profit maximizing price:

Δπ / ΔP =12− 4P + P = 0
1 1 1 2

Firm 1’s reaction curve:

P1 = 3 + 1 P2
4
Firm 2’s reaction curve:
P2 = 3 + 1 P1
4
Nash Equilibrium in Prices
Firm 1’s reaction curve gives its profit-
maximizing price as a function of the price
that Firm 2 sets, and similarly for Firm 2.
The Nash equilibrium is at the intersection
of the two reaction curves: When each firm
charges a price of $4, it is doing the best
it can given its competitor’s price and has
no incentive to change price.
.

Bertrand-Nash equilibrium (for

Differentiated Product Duopoly)
P1 = P2 = $4
q1 = q2 = 8, π1 = π2 = 12
Dynamic (Sequential) Oligopoly
Ø  Firms may make decisions sequentially, a firm moving first and the other firms
deciding after the first firm decided.

Ø  Two varieties:
Ø  Cournot-Stackelberg Model:
─  Firms choose quantities sequentially. (Original Stackelberg model)

Ø  Bertrand-Stackelberg Model:
─  Firms choose prices sequentially.
Cournot – Stackelberg Duopoly
Ø  Two firms – firm 1 and firm 2 – producing homogeneous products.

Ø  Market Demand: P = 30 – Q = 30 – (q1 + q2)

Ø  FC = VC = 0.(i.e. MC1 = MC2 = 0)

Ø  Firm 1 acts as a leader, choosing its quantity first, and firm 2 is a follower. The
follower firm is in the same situation as a Cournot firm: it takes the leader’s output as
given and maximizes profits accordingly.

Ø  The leader chooses the level of output that maximizes its profits given the follower
reacts to what the leader does.
What is the most likely outcome?
o  Two firms – firm 1 and firm 2 – producing homogeneous products.
o  Market Demand: P = 30 – Q = 30 – (q1 + q2)
o  FC = VC = 0.(i.e. MC1 = MC2 = 0)

§  Firm 2 (2nd mover or follower):

would follow its best response/reaction
q2 = 15 – (1/2)q1 (from the previous slide)

§  Firm 1 (1st mover or leader): q2
Firm 1’s
π1 = TR1 – TC1 = (30 – q1 – q2)q1 Best Response
= [30 – q1 – (15 – (1/2)q1)]q1 = 15q1 – 1/2q 2
1
dπ1/dQ1 = 15 – q1 = 0
Cournot-Nash
Thus, q1 = 15
Cournot-Stackelberg (with
firm 1 being the leader and
q  Cournot-Stackelberg Equilibrium: firm 2 the follower)

q1 = 15, q2 = 7.5, Q = 22.5, 10
P = $7.5, Firm 2’s
π1 = 112.5 π2 = 56.25 7.5 Best Response

q  Remarks: First mover advantage exists. q1
(Play ‘Top Dog’) 10 15

8
Bertrand – Stackelberg Duopoly with
Homogeneous Products
A Model:
─  Two firms – firm 1 and firm 2 – producing homogeneous products yet setting
the prices sequentially.
─  Market Demand: Q = 30 – P
(Assume: Lower-pricing firm takes the whole market.
If equal price, two firms split the market by halves.)
─  MC1 = MC2 = $3.

q  Subgame Perfect Nash Equilibrium (SPNE) Strategies

ü  Firm 2: P2 = P1 – $0.01 as long as P1 > MC1 = 3. Otherwise, P2 =3.

ü  Firm 1: Given Firm 2’s reaction, it would charge P1= 3.

q  SPNE Outcome

ü  Both firms charge P=$3 and only get competitive outcome.

q  Remarks: No first or second mover advantage exists. (Note that the outcome is the
same as under static Bertrand Nash equilibrium.)
9
Strategic Implications Wrap-Up
Ø  For Quantity Competition
q(leader) > q(follower)
π(leader) > π(follower)
That is, the first mover advantage exists.

Ø  For Price Competition (with Homogeneous Products)

P(follower) = P(leader)
π(follower) = π(leader) = 0
That is, no first or second mover advantage exists.

10
II. Game Theory I
 Outline
Ø  Introduction, Motivating Examples and Fundamentals of a Game

Ø  Strategic Thinking in Static Games

─  Some Equilibrium Concepts
o  Dominant Strategy Equilibrium

o  Iterated Dominance Equilibrium

o  Maximin Strategy Equilibrium
How to Buy a Dollar Bill
Ø  Consider the game devised by Martin Shubik. A dollar bill is auctioned, but in an
unusual way. The highest bidder receives the dollar in return for the amount bid.
However, the second-highest bidder must also hand over the amount that he or she
bid—and get nothing in return.

Ø  If you were playing this game, how much would you bid for the dollar bill?

Ø  Mutual interdependence in decision-making process is the key aspect of game

situation. Thus strategic thinking is the focus of game theory.
 Introduction
Ø  Game theory is the branch of microeconomics concerned with the analysis of optimal
decision making in competitive situations, in which the actions of each decision maker
have a significant impact on the fortunes of rival decision makers. Such situations are
common both within firms and in market actions among firms as well as in everyday
life.

─  Capacity competition (e.g. collusion vs competition?) or Price wars

─  You broke up with your girl (boy) friend a week ago. And you want her
(him) back. Should you call ahead or wait for her (his) call? What is the
winning strategy? More importantly, what matters in your decision?

Ø  In these situations, players care their own interests only– possibly in conflict – but
also share a mutual dependence.
Example: Honda & Toyota
Ø  The market for automobiles in china experienced a boom during the first decade of
the new millennium. By 2009 the number of light vehicles sold in China had equaled
the number sold in the US.
Ø  Major automobile firms like Honda and Toyota often relish the opportunity to enter
growing marketing around world.
Ø  Besides the growth in demand, what else should they take into account?
Ø  Profitability depends on many factors, such as Market Demand, Production Cost,
Market Supply, and Decisions Made by Rival Firms.
Ø  For example, in the late 1990s, both Honda and Toyota had to decide whether to
build new auto assembly plants in North America.
Toyota
Build a new plant Don’t build a new
plant
Honda

Build a new plant 16, 16 20, 15

Don’t build a new 15, 20 18, 18

plant
Gaming and Strategic Decisions
•  game Situation in which players (participants) make strategic decisions
that take into account each other’s actions and responses.

•  players A set of rational, payoff-maximizing competitors.

•  payoff Value associated with a possible outcome.

•  strategy Rule or plan of actions for playing a game.

•  best response The strategy of one player that results in the best payoffs to him/
her, given the combination of other players’ strategies.

If I believe that my competitors are rational and act to maximize their own payoffs,
how should I take their behavior into account when making my decisions?
Noncooperative versus
Cooperative Games
•  cooperative game Game in which participants can negotiate
binding contracts that allow them to plan joint strategies.

•  noncooperative game Game in which negotiation and enforcement of binding

contracts are not possible.

It is essential to understand your opponent’s point of view and to deduce his or her likely
responses to your actions.

Note that the fundamental difference between cooperative and noncooperative

games lies in the contracting possibilities. In cooperative games, binding contracts are
possible; in noncooperative games, they are not. Since collusion in many countries are
illegal, we only study non-cooperative games.
How to classify a game?
Ø  The sequence of moves: sequential or simultaneous?

Ø  Are the players' interests in total conflicts or is there any

commonality?

Ø  Is the game played once or repeatedly?

Ø  Do the players have full or equal information?

What is the likely outcome of a game?
Ø  Game theory predicts optimal strategy for each player.
•  The optimal strategy maximizes a player’s payoffs given others’
strategic plays.
•  Game theory predicts how the game is going to be played in
obviously reasonable ways.

Ø  Solution concept: Nash Equilibrium

•  Nash Equilibrium is a strategy profile (a combination of strategy),
where no player in the game has the incentive to deviate.
•  In a Nash Equilibrium, each player's strategy is the best response to
other players' strategies. A player is unable to do strictly better by
unilaterally switching his/her strategy.
Some other solution concepts
Ø  Pre-Nash Equilibrium Concepts
§  Dominant Strategy Equilibrium
§  Iterated (or Iterative) Dominance Equilibrium
§  Maximin Strategy Equilibrium

Ø  Post-Nash Equilibrium Concept

§  Subgame-Perfect Nash Equilibrium
One-Shot, Simultaneous-Move Games
Ø  We use normal-form to represent the game.

§  Players simultaneously decide their strategies.

§  A representation of a game indicating the players, their possible strategies, and

the payoffs resulting from alternative strategies.

§  One player chooses strategy from the “row”, while the other player chooses
strategy from the “column”.

§  In each cell, the first entry indicates the payoff of the row player, while the second
entry indicates the payoff of the column player.
 Dominant Strategy Equilibrium
● dominant strategy Strategy that is optimal no matter what an opponent does.

TABLE 13.1 PAYOFF MATRIX FOR ADVERTISING GAME

Firm B
Advertise Don’t advertise

Advertise 10, 5 15, 0

Firm A
Don’t advertise 6, 8 10, 2

Advertising is a dominant strategy for Firm A. The same is true for

Firm B: No matter what firm A does, Firm B does best by advertising. The outcome for
this game is that both firms will advertise.
Formalization:
Concept of Dominance
Ø  The strategy s* is a dominant strategy if it is a player’s optimal choice no matter
what the other players do.

Ø  His inferior strategies are called dominated strategies, which a rational player
would not choose.

Ø  When every player has a strictly dominant strategy, the outcome of the game is
called a dominant-strategy equilibrium.
 

Example:
In the game, “Advertise’ becomes a dominant strategy for both players.
(Advertise, Advertise) shares such property for both players, and thus
constitutes a dominant strategy equilibrium.
Example for Dominant Strategy Equilibrium:
Pricing Game
•  Recall the fruit price war in Ang Mo Kio (Singapore).
─  Two shops lost $30,000 and $50,000 over 5 days of price war.

•  ‘All pricing high’ would be

the best for both.
But,
•  ‘All Pricing Low’ is a
dominant strategy
equilibrium.

Our goals

•  Learn how to avoid it from the beginning.

•  Once caught, learn how to get out of it.

24
Unfortunately, not every game has a dominant strategy for each player.

TABLE 13.2 MODIFIED ADVERTISING GAME

Firm 2
Advertise Don’t advertise

Advertise 10, 5 15, 0

Firm 1
Don’t advertise 6, 8 20, 2

Now Firm A has no dominant strategy. Its optimal decision depends on what
Firm B does. If Firm B advertises, Firm A does best by advertising; but if Firm B
does not advertise, Firm A also does best by not advertising.
Iterated Dominant Equilibrium
Ø  Dominant strategy equilibrium was about ‘which strategy will be played for sure?’
 
•  If the dominant strategy equilibrium exists, it is easy to solve.
•  But few games have the dominant strategy equilibrium.
 
Ø  Then what if there is no dominant strategy equilibrium?

•  Then, as an alternative, we can ask ‘what strategies will not be played?’

̶  The rationale behind is clear: you wouldn’t play a strategy when there is some
other strategy which always does better regardless of the rival’s choices.
̶  We call such a strategy a dominated strategy.

If dominated, you won’t adopt that strategy and you would delete it out of your
option. Iteratively deleting such dominated strategies can – although not always
– give us a unique outcome: it is called ‘iterated dominance equilibrium’.

26
Simplify the game by successive
elimination of dominated strategies
Toyota

Build large Build small Don’t build

Honda

Build large 0, 0 12, 8 18, 9

Build small 8, 12 16, 16 20, 15

Don’t build 9, 18 15, 20 18, 18

In this game, no matter what player Toyota does, the strategy “Build Large” results
in strictly worse payoffs than “Build Small” for Honda. Then delete it!

Toyota

Build large Build small Don’t build

Honda

Build large 0, 0 12, 8 18, 9

Build small 8, 12 16, 16 20, 15

Don’t build 9, 18 15, 20 18, 18

In the revised game, no matter what Honda does, the strategy “Build Large” is
strictly dominated by both “Build Small” and “Don’t Build” for Toyota. Then delete
it!
Toyota

Build large Build small Don’t build

Honda

Build large 0, 0 12, 8 18, 9

Build small 8, 12 16, 16 20, 15

Don’t build 9, 18 15, 20 18, 18

Iterate the above procedure until you cannot proceed further

Toyota

Build small Don’t build

Honda

Build small 16, 16 20, 15

Don’t build 15, 20 18, 18

(Build small, Build small) is the only outcome that survives the ‘Iterated Dominance’,
thus iterated dominance equilibrium.

(The deletion process does not always yield the unique outcome – which is why this equilibrium
selection algorithm is not universal enough.)
Maxmin Strategy Equilibrium
Ø  Players may sometimes play conservatively to avoid a bad outcome. Such risk-averse
players may suspect that other players are not necessarily rational. Then they would be
willing to tradeoff between risk and return. In doing so, their objective is to choose the
strategy associated with the best out of the worst outcomes.

A strategy of Player A that guarantees herself the best outcome among the worst
possible outcomes that arise from her strategy choices – that is, the maximum out of
the minimum payoffs – is called A’s maximin strategy.

31
Maximin Strategies
TABLE 13.4 MAXIMIN STRATEGY
Firm 2
Firm 1’min
Don’t invest Invest

Don’t invest 0, 0 –10, 10 -10

Firm 1
Invest –100, 0 20, 10 -100

Firm 2’min 0 10

In this game, the outcome (invest, invest) is the most likely outcome if both firms are
rational players. But if you are concerned that the managers of Firm 2 might not be fully
informed or rational—you might choose to play “don’t invest.” In that case, the worst that
can happen is that you will lose $10 million; you no longer have a chance of losing
$100 million. Hence, (Don’t invest, invest) is a Maxmin strategy equilibrium.

● maximin strategy Strategy that maximizes the minimum gain that can be earned.
 Maximin Strategy Equilibrium– cont.

Skepticisms of Maximin Strategy Concept:

 
•  Very little justification for an equilibrium concept for a rational player:
-  The concept may sound practical, particularly when the damage from the
bad outcome is fatal. But how bad is fatal? Would you still behave
conservatively when (Invest, Don’t invest) gives (-1, 0) instead of (-100, 0)?

•  In sum, this concept may be acceptable for a risk-averse behavior but inconsistent
and fragile as an equilibrium concept.