HANDOUT THREE: GAME THEORY

Introduction
The subject matter of game theory is to study strategic decisions characterized by interactive
payoffs. It is concerned with choice of the best or optimal strategy in conflict situations.

Each individual’s welfare depends not only on his / her actions but also on the actions of other
individuals. An oligopolistic firm wishing to maximize profits by changing the product price, the
quantity of the product it sells, the level of advertising, etc. must take into consideration all the
possible reactions of its competitors. Every action taken by the firm and the reaction of the
competitors will result in different payoffs or outcomes. Under strategic decisions, the optimal
choice of the firm may change depending on what it believes its competitors will do.

A game is a way to model strategic behavior – people recognize that their own behavior affects
the choices of others and that the outcome depends on any one person’s choice, not just one’s
own Strategic behavior is important in situations:
1. Where a small number of individuals interact (negotiations, auctions)
2. There is imperfect competition
Game theory not limited to economics e.g. biology, sociology, political science, dating, sports, ...

Basic Concepts in Game Theory

A game is a formal representation of a situation in which a number of individuals interact in a
setting of strategic interdependence.

Strategic interdependence: The optimal choice of one player depends on what his rival chooses.
The action of one player affects the welfare of the other. Therefore, when a player is making the
decision on his best strategy, he / she must take account of the possible actions of the other
player in order to maximize his / her payoff.

Player: The decision makers whose behavior the game predicts.

Rules: Describe the order of play and the information available to each player when it is their
turn to act.

Feasible strategy set: A complete contingent plan describing the possible actions the player can
take at every possible situation that they may be called upon to move.

Strategy: Is a complete contingent plan that specifies how the players will act in every possible
circumstance in which he / she might be called upon to act.
Dominant strategy: A strategy is said to be dominant for a player if it maximizes the players
payoff for any strategy that the rival might play.

The payoff matrix: Strategic interaction can involve many players and many strategies, but we’ll
limit ourselves to two-person games with a finite number of strategies. This will allow us to
depict the game easily in a payoff matrix. It is simplest to examine this in the context of a
specific example. Suppose that two people are playing a simple game. Person A will write one of
two words on a piece of paper, “top” or “bottom.” Simultaneously, person B will independently
write “up” or “down” on a piece of paper.

Illustration:

Top Up
Player 1: Strategies Bottom Player 2: Strategies Down

Given payoffs, the game can be represented in two forms;

1. Extensive / tree form

3,0
U

T [ 2] D 3 ,−2
[ 1]
B [ 2] U 1,−1
 D
4 ,2

2. Normal/ Matrix form

Player 2
U D
T 3,0 3 ,−2
Player1 B 1,−1 4 ,2

Consider a market with two firms, Firm A and Firm B. The firms seek to increase sales through
advertising, therefore have two strategies, Advertise or Don’t advertise. Each firm expects to
increase profits by advertising though the actual level of profit depends on whether the rival
advertises or not. Therefore, each firm’s strategy depends with the rival’s strategy. The possible
outcomes and payoffs are illustrated in the following payoff matrix.

Firm B

Advertise Don’t Advertise

Firm A Advertise 4, 3 5, 1

Don’t Advertise 2, 5 3, 2

How to read the payoff matrix: From the table, if both firms do not advertise Firm A will earn a
profit of 3, and Firm B will earn a profit of 2 (bottom right cell of the payoff matrix).
The top right cell of the payoff matrix shows that if Firm A advertises and Firm B does not, Firm
A will have a profit of 5, and firm B will have a profit of 1.

The above game can also be presented in extensive form (game tree) showing who moves when,
what action (strategy) each player can take, what players know when it is their turn to move,
what the outcome is as a function of the actions taken by the players, and the player’s payoffs
from each possible outcome.
SEQUENTIAL MOVE GAMES
In many situations one player gets to move first, and the other player responds. Sequential games
are played by the players taking turns (moves alternate between players). In presentation there is
no information set implying player 2 knows what player 1 has played. The game tree of a
sequential move game of the advertising assuming player A makes the first move of either
advertising or not is presented below. After seeing Firm A’s choice, Firm B makes a move either
of advertising or not advertising.
.
Firm A

Advertise Don’t Advertise

Firm B Firm B

Advertise Don’t Advertise Don’t Advertise

Advertise

Firm A’s payoff 4 5 2 3

Firm B’s payoff 3 1 5 2

We assume that Firm B knows Firm A’s move when it acts.

The game starts at an initial decision node represented by an open square, where Firm A makes
it’s move, deciding whether to advertise or not to advertise. Each of the two possible choices for
Firm A is represented by a branch from this decision node. At the end of each branch is another
decision node represented by a solid square, at which Firm B can choose to advertise or not, after
seeing Firm A’s choice. After Firm B’s move we reach the end of the game, represented by
terminal nodes. At each node, we list the player’s payoffs arising from the sequence of moves
leading to that node.

Exercise

In the first round, player A gets to choose top or bottom. Player B gets to observe the first
player’s choice and then chooses left or right. The payoffs are illustrated in a game matrix

Player B

Left Right

Player A Top 1, 9 1, 9

Bottom 0, 0 2, 1

Present the above as a game tree assuming that the players know the strategy that the competitor has
chosen.

SIMULTANEOUS MOVE GAME

In a simultaneous move game there is imperfect information. When it is a player’s turn to make a
move he / she does not know which decision node the other player is at (the action of the other
player). The player has no opportunity to observe what has happened in the game. A circle is
drawn around player 2’s decision nodes to indicate this as shown in the following coin tossing
game. However, Player 1 likewise does not know player 2’s action.

Player 1

Head Tail

Player 2
 Head Tail Head Tail

Player 1’s payoff -5 5 5 -5

Player 2’s payoff 5 -5 -5 5

Exercise: Represent the above game in matrix form.

Example Two: Both players have no information at all, i.e. player 2 is in the information set.
3,0
U

T [ 2] D 3 ,−2
[ 1]
B [ 2] U 1,−1
D
4 ,2
Information set

Solving Games

1. Dominant Strategy Equilibrium/ Successive elimination of Dominated Strategies

Dominant strategy is one which is optimal regardless of the opponent’s action – one that is
always chosen. We eliminate the dominated strategies.

Example; 2 firms decide simultaneously whether to advertise or not.

Firm 2
Advertise Don’t Advertise
Advertise 10 , 5 13 , 2
Firm 1 Don’t Advertise 6,7 11, 9
“Advertise” is a dominant strategy for firm 1 hence “Don’t Advertise” is eliminated. Since both
firms know that firm 1 will advertise, what will firm 2 do? It will advertise.

Therefore, solution is(10 , 5) , we don’t report the pay offs (numbers), we report the strategies; so
equilibrium is “advertise, advertise”.

Example 2

Two firms, 1 and 2 produce different products with payoffs as follows.

Firm 2

L C R

T 3, 0 2,1 0,0

Firm 1 B 1,1 1,1 5,0

D 0,1 4,2 0,1

Using successive elimination of dominated strategy, solve the game.

Example 3

Player 2

L R

Player 1 T 20 ,−20 −30 , 30

B −20 , 20 10 ,−10

This is another zero – sum game. No strategy can be eliminated hence we use the Nash
equilibrium application to solve.
2. Nash Equilibrium
Dominant strategy equilibria are nice when they happen, but they don’t always happen. For
example, the game depicted in the following table does not have a dominant strategy.

Player B

Left Right

Player A Top 2, 1 0, 0

Bottom 0, 0 1, 2

Here when B chooses left the payoffs to A are 2 or 0. When B chooses right, the payoffs to A are
0 or 1. This means that when B chooses left, A would want to choose top; and when B chooses
right, A would want to choose bottom. Thus A’s optimal choice depends on what he thinks B
will do.

We will say that a pair of strategies is a Nash equilibrium if A’s choice is optimal, given B’s
choice, and B’s choice is optimal given A’s choice. Remember that neither person knows what
the other person will do when he has to make his own choice of strategy. But each person may
have some expectation about what the other person’s choice will be.

A Nash equilibrium can be interpreted as a pair of expectations about each person’s choice such
that, when the other person’s choice is revealed, neither individual wants to change his behavior.
In the above table if A chooses top, then the best thing for B to do is to choose left, since the
payoff to B from choosing left is 1 and from choosing right is 0. And if B chooses left, then the
best thing for A to do is to choose top since then A will get a payoff of 2 rather than of 0. Thus if
A chooses top, the optimal choice for B is to choose left; and if B chooses left, then the optimal
choice for A is top. So we have a Nash equilibrium: each person is making the optimal choice,
given the other person’s choice. The strategy (top, left) is a Nash equilibrium. The weakness of
this application is that it may give more than 1 equilibrium, it may not exist, and it may not
necessarily lead to Pareto efficient outcome.

Example 1 Firm 2

L R

Firm 1 T 20, -20 -30, 30

B -20, 20 10, -10

No Nash equilibrium

Example 2 Boeing

Dev D.D

Airbus Dev -10, -10 5, -1

D.D -1, 5 0, 0

We have 2 Nash equilibriums, i.e. ital Dev, ital DD or ital DD, ital Dev

Example 3

Prisoners’ dilemma

The Nash equilibrium of a game does not necessarily lead to Pareto efficient outcomes. The
Prisoner’s Dilemma depicts both Nash equilibrium as well as dominant strategy equilibrium, but
is not socially optimal.
The original discussion of the game considered a situation where two prisoners who were
partners in a crime were being questioned in separate rooms. Each prisoner had a choice of
confessing to the crime, and thereby implicating the other, or denying that he had participated in
the crime. If only one prisoner confessed, then he would go free, and the authorities would
require the other prisoner to spend 15 months in prison. If both prisoners denied being involved,
then both would be held for 1 month on a technicality, and if both prisoners confessed they
would both be held for 10 months. The payoff matrix for this game is as follows:

Prisoner 2
Confess Don’t Confess
Confess −10 ,−10 0 ,−15
Prisoner 1 Don’t Confess −15 , 0 −1,−1

In this game, cheating is optimal hence players don’t corporate even though corporation would
be socially optimal (Pareto Improving). All games that depict such scenario are referred to as
prisoners’ dilemma.

How can the players be made to corporate?

1. Appointing an authoritative representative body to keep, check and penalize.

2. The game be played repeatedly (i.e. dynamic games).
What if there is no Nash equilibrium? And no dominant strategy equilibrium in pure strategies?

3. Mixed Strategies
If we enlarge our definition of strategies, we can find a new sort of Nash equilibrium. We have
been thinking of each agent as choosing a strategy once and for all. That is, each agent is making
one choice and sticking to it. This is called a pure strategy.
A game with no Nash equilibrium (in pure strategies)

Player B

Left Right

Player A Top 0,0 0 , -1

Bottom 1,0 -1 , 3

If we allow the agents to randomize their strategies therefore assign a probability to each choice,
and to play their choices according to those probabilities. For example, A might choose to play
top 50 percent of the time and bottom 50 percent of the time, while B might choose to play left
50 percent of the time and right 50 percent of the time. This kind of strategy is called a mixed
strategy.

If A and B follow the mixed strategies given above, of playing each of their choices half the
time, then they will have a probability of 1/4 of ending up in each of the four cells in the payoff
matrix. Thus the average payoff to A will be 0, and the average payoff to B will be 1/2. A Nash
equilibrium in mixed strategies refers to an equilibrium in which each agent chooses the optimal
frequency with which to play his strategies given the frequency choices of the other agent.

There will always exist a Nash equilibrium in mixed strategies. Because a Nash equilibrium in
mixed strategies always exists, and because the concept has a certain inherent plausibility, it is a
very popular equilibrium notion in analyzing game behavior. In the example it can be shown that
if player A plays top with probability 3/4 and bottom with probability 1/4, and player B plays left
with probability 1/2 and right with probability 1/2, this will constitute a Nash equilibrium.

We use mixed strategies where probabilities are assigned to the strategies and then we obtain the
players’ expected payoffs.
Question
Two competing firms intend to launch new products in the market. Firm 1 has three products, d,
e and f, while Firm 2 has products a, b, and c. The payoff matrix of the product launching game
is as below:
Firm 2

Product a Product b Product c

Product d 3, 6 7, 1 10, 4

Firm 1 Product e 5, 1 8, 2 14, 7

Product f 6, 0 6, 2 8, 5

Assuming that both players are rational therefore will never play a dominated strategy, find the
equilibrium of the game using iterative elimination of dominated strategies.

Question
The following is the payoff matrix of player 1 and player 2.

Player 2

Product A Product B Product C

Product D 3, 6 7, 1 10, 4

Player 1 Product E 5, 1 8, 2 14, 7

Product F 6, 0 6, 2 8, 5

Assuming that both players are rational therefore will never play a dominated strategy:
Find the equilibrium of the game using iterative elimination of dominated strategies.
Question
 Firm B

Advertise Don’t Advertise

Firm A Advertise 4, 3 5, 1

Don’t Advertise 2, 5 3, 2

Question: Does Firm A have a dominant strategy in the above game.

What is the equilibrium of this game?