ECO 401 AND 411

TOPIC FOUR
THEORY OF GAMES

RS DAUDA
CONTENTS
 Introduction
 Description of a Game
 Some Important Concepts in Games Theory
 Assumptions Made in Games Theory
 Methods for or steps in solving game problems
 Description and Types of Games
 Two-Person Zero-Sum Game:
 Pure Strategies
 Dominating Strategies
 Mixed Strategies
 Game Theory using decision tree
 References
 INTRODUCTION
 Life is full of conflict and competition, which cut across households,
firms and government.

 In taking decisions at these levels conflicts and completions are bound

to occur.

 Game theory is a mathematical theory that deals with the general

features of competitive situations.

 This theory is helpful when two or more individuals or organisations

with conflicting objectives try to make decisions.

 Thus, Game theory deals with decision situations in which two

intelligent opponents with conflicting objectives are vying to outdo one
another.

 In such a situation, a decision made by one person affects the decision

made by one or more of the remaining decision makers, and the final
outcome depends on the decision of all the parties.

 Because competitive situations are so ubiquitous, game theory has

applications in a variety of areas/fields, such as economics, business,
military, etc.
INTRODUCTION . . .
 In economics. the theory of non-cooperative games studies and models
conflict situations among economic agents.
 It studies situations where the profits (gains, utility or payoffs) of each
economic agent depend not only on her own acts but also on the acts of the
other agents.

 Because players are rational, so each will try to maximize her profit function
(utility or payoff) given her conjectures or beliefs on how the other players
are going to play.

 The outcome of the game will depend on the acts of all the players.
 A fundamental characteristic of non-cooperative games is that it is not
possible to sign contracts between players.

 That is, there is no external institution (for example, courts of justice)

capable of enforcing the agreements.

 In this context, co-operation among players only arises as an equilibrium or

solution proposal if the players find it in their best interest.

 von Neumann and Morgenstern were the first to employ games approach
systematically solve oligopoly problem in their book titled: “Theory of Games
and Economic Behaviour”, published in 1944.
 DESCRIPTION OF A GAME
 Our focus here is on competition between two firms, which has more earning
competitive situations.

 In this topic, a competitive situation is called a game if it has the following

characteristics (Assumptions made to define it):

 1. There is finite number of competitors called Players.

 The game they play may be for creating new market, increasing the market share
or increasing the competitiveness of the product.

 2. A play is played when each player chooses one of his courses of actions. The
choices are made simultaneously, so that no player knows his opponent's choice
until he has decided his own course of action.

 Zero sum game (ZSG): This is a game in which the gains of one player are the
losses of other player or the algebraic sum of gains of both players is equal to
zero).

 Thus, in a zero sum game, the algebraic sum of the gains of all players after play
is bound to be zero.

 A zero sum game is also known as constant sum game.

 A non zero-sum game: This is a game in which the sum of gains and/or losses is
not equal to zero.
 DESCRIPTION OF A GAME…
 In game theory, the resulting gains can easily be represented
in the form of a matrix called pay–off matrix or gain matrix.

 A pay-off matrix is a table, which shows how payments should

be made at end of a play or the game.

 A zero sum game involving two persons is called a Two-Person

Zero-Sum Game (TPZSG).

 A good example of two- person game is the game of chess.

 The activities of duopolist can also come under two- person

game.

 A good example of n-person game is the situation where

several companies are engaged in an intensive advertising
campaign to capture a larger share of the market (Murthy,
2007)
SOME IMPORTANT CNONCEPTS IN GAMES THEORY
 1) Player: A player is an active participant in a game. The games can
have two persons (two-person game) or more than two persons (multi
person or n-person game)
 2) Game: A game is a sequence of moves that are defined by a set of
rules that governs the players‟ moves. The sequence of moves may be
simultaneous.

 3) Decision maker: A decision-maker is a person or group of people

in a committee who makes the final choice among the alternatives.
With respect to game, a decision-maker is the player in the game.

 4) Moves: A move could be a decision by player or the result of a

chance event.

 5)Objective: An objective is what a decision-maker aims at

accomplishing by means of his decision and this could be more than
one.
 6) Behaviour: This could be any sequence of states in a system.

 The behaviour of a system is overt while state trajectories are covert.

 SOME IMPORTANT COMCEPTS IN GAMES THEORY…
 7) Strategy: The strategy for a player is the list of all possible actions (moves, decision
alternatives or courses of action) that are likely to be adopted by him for every payoff
(outcome).
 It is a specific course of action with clearly defined values for the policy variables. For
example, a strategy may consist of setting a price of N300, spending N200000 on advertising,
making a change in the packaging of the product, and selling it in supermarkets.

 It is assumed that the players are aware of the rules of the game governing their decision
alternatives (or strategies).

 The outcome resulting from a particular strategy is also known to the players in advance and
is expressed in terms of numerical values (e.g. money, per cent of market share or utility).

 The particular strategy that optimizes a player‟s gains or losses, without knowing the
competitor‟s strategies, is called optimal strategy.

 The expected outcome when players use their optimal strategy, is called value of the game.

 Generally, the following two types of strategies are followed by players in a game:

 (a) Pure Strategy: A particular strategy that a player chooses to play again and again
regardless of other player‟s strategy is referred as pure strategy. The objective of the players
is to maximize their gains or minimize their losses.

 (b) Mixed Strategy: A set of strategies that a player chooses on a particular move of the game
with some fixed probabilities.

 Thus, there is a probabilistic situation and objective of the each player is to maximize
expected gain or to minimize expected loss by making the choice among pure strategies with
fixed probabilities.
 SOME IMPORTANT DEFINITIONS IN GAMES THEORY…
 8)Decision: The forceful imposition of a constraint on a set of initially possible
alternatives.
 9) Conflict: A condition in which two or more parties claim possession of
something they cannot all have simultaneously.

 10) Perfect information: A game is said to have perfect information if at every

move in the game all players know the move that have already been made. This
includes any random outcomes.

 11) Zero-sum game: A game is said to be zero sum if the sum of player‟s payoff is
zero. The zero value is obtained by treating losses as negatives and adding up the
wins and the losses in the game.

 12) Payoffs: This is the numerical return received by a player at the end of a
game and this return is associated with each combination of action taken by the
player.

 The payoff of a strategy is the same as the 'net gain' it will bring to the firm for
any given counterstrategy of the competitor(s). This gain is measured in terms of
the goal(s) of the firm.
 For example, if the goal of the firm is to maximise its profit, the payoffs of a
strategy will be measured in terms of profit levels that it yields; if the goal is
maximisation of the market share, the payoffs will be measured as the actual
shares that the strategy will secure to the firm adopting it.

 We talk of “expected payoff‟ if its move has a random outcome.

 SOME IMPORTANT DEFINITIONS IN GAMES THEORY…
 13) Payoff matrix: The payoff matrix of a firm is a table showing the
payoffs accruing to this firm as a result of each possible combination of
strategies adopted by it and by its rival(s).

 For example, assume that there are two firms in an industry.

 Firm I has to choose among five strategies (A1.A2 ,•••,A5) and Firm II
can react by adopting any one of six strategies open to it (B1,B2 ,•••,B6).

 Thus to each strategy of Firm I there are six possible counterstrategies of

Firm II, and similarly to each strategy of Firm II there are five
counterstrategies of the rival Firm I.

 Thus the payoff matrix of each firm will include 5 x 6 = 30 payoffs,

corresponding to the results of each possible combination of strategies
selected by both rivals.

 Let us denote each payoff as Gij, where i refers to the strategy adopted by
Firm I and j to the counter-strategy adopted by Firm II. Thus for the
above example the payoff matrix for Firm I will be of the general form of
Table 4.1.
 TABLE 4.1: PAYOFF MATRIX OF FIRM I
Firm II's strategies
Firm I's B1 B2 B3 B4 B5 B6
strategies
A1 G11 G12 G13 G14 G15 G16
A2 G21 G22 G23 G24 G25 G26
A3 G31 G32 G33 G34 G35 G36
A4 G41 G42 G43 G44 G45 G46
A5 G51 G52 G53 G54 G55 G56
From the Table, if Firm I adopts strategy A1 and its rival reacts by adopting
among the strategies open to it like B5, the payoff (gain) of Firm I will be G15 . If
Firm I chooses strategy A4 and its rival reacts with strategy B6, the payoff of
Firm I will be G46, and so on.
In the theory of games the firms in say oligopolistic markets are treated as
players in a chess game: to each movement by one player the other may choose
among several counter-movements.
The counter-movements of rivals are probable but not certain. Yet it is possible
to choose a strategy which (under certain conditions) will maximise the firm's
expected 'gain', after making due allowance for the effects of rivals' probable
reactions.
 ASSUMPTIONS MADE IN GAMES THEORY
 The following are assumptions made in games theory.

 1)Each player (decision-maker) has available to him two or more clearly specified
choices or sequence of choices (plays).

 2)A game usually leads to a well-defined end-state that terminates the game. The
end state could be a win, a loss or a draw.

 3)Simultaneous decisions by players are assumed in all games.

 4) A specified payoff for each player is associated with an end state (eg sum of
payoffs for zero sum-games is zero in every end-state).

 5) Repetition is assumed. A series of repetitive decisions or plays results in a

game

 6) Each decision-maker (player) has perfect knowledge of the game and of his
opposition, i.e. he knows the rules of the game in details and also the payoffs of
all other players.

 The cost of collecting or knowing this information is not considered in game

theory.

 7) All decision-makers are rational and will therefore always select among
alternatives, the alternative that gives him the greater payoff.
METHODS FOR OR STEPS IN SOLVING GAME PROBLEMS
 Step 1. Find the Saddle point: If saddle point exists, then the
game can be solved. All you need to do is to write the optimal
strategies and the value of the game.

 Step 2. If there is no saddle point, you can reduce the matrix

to: 2 x 2 or 2 x 3 or 3 x 2 as the case may be and use the
appropriate method to solve the problem.

 Step 3. You can use the algebraic method to get the solution.

 Step 4. You could use the linear programming to get the

solution.

 Step 5. Iteration method can also be used to determine the

solution.
DESCRIPTION AND TYPES OF GAMES
 Games can be described in terms of the number of players
and the type of sum obtained for each set of strategies
employed. Thus, we have the following types of games:

 1) Two-person zero-sum games. Here two players are

involved and the sum of the pay-offs for every set of
strategies by the two players is zero

 2) Two-person non zero-sum games. Here two players are

involved and there is one strategy set for which the sum of
the payoffs is not equal to zero.

 3) Non-constant sum games. The values of payoffs for this

game vary.

 4) Multi-person non-constant-sum games. Many players are

involved in the game and the payoffs for the players vary.
 TWO-PERSON ZERO-SUM GAME
 This game involves two players in which losses are treated as negatives and wins as
positives and the sum of the wins and losses for each set of strategies in the game is
zero.

 Whatever player one wins player two loses and vice versa.

 Each player seeks to select a strategy that will maximise his payoffs although he does
not know what his intelligent opponent will do.

 A two-person zero-sum game with one move for each player is called a rectangular
game.

 Characteristics of Two-Person Zero-Sum Games

 1 There are two players (called the row player and the column player).
 2 The row player must choose 1 of m strategies.
 Simultaneously, the column player must choose 1 of n strategies.
 3 If the row player chooses his ith strategy and the column player chooses his jth
strategy, then the row player receives a reward of aij and the column player loses an
amount aij.
 Thus, we may think of the row player‟s reward of aij as coming from the column
player.
 A two-person zero-sum game is represented by the matrix in Table 4.2 (the game‟s
reward matrix).
 TABLE 4.2: EXAMPLE OF TWO-PERSON ZERO-SUM GAME
Row Player’s Column Player’s Strategy
Strategy
Column 1 Column 2 … Column n

Row 1 a11 a12 … a1n

Row 2 a21 a22 … a2n

. . . … .
. . . .
. .
. .

Row m am1 am2 … amn

As previously stated, aij is the row player‟s reward (and

the column player‟s loss) if the row player chooses his ith
strategy and the column player chooses his jth column
strategy.
TWO-PERSON ZERO-SUM GAME…
 For example, in the two-person zero-sum game in Table
4.3, the row player would receive two units (and the
column player would lose two units) if the row player
chooses his second strategy and the column player
chooses his first strategy.

 A two-person zero-sum game has the property that for

any choice of strategies, the sum of the rewards to the
players is zero.

 In a zero-sum game, every naira that one player wins

comes out of the other player‟s pocket, so the two
players have totally conflicting interests.

 Thus, cooperation between the two players would not

occur.
TABLE 4.3: EXAMPLE OF TWO-PERSON ZERO-SUM GAME
Row Player’s Column Player’s Strategy
Strategy
Column 1 Column 2 Column 3 Column 4

Row 1 1 2 3 -1

Row 2 2 1 -2 0

The row player would receive two units

(and the column player would lose two
units) if the row player chooses his
second strategy and the column player
chooses his first strategy.
BASIC ASSUMPTION OF TWO-PERSON ZERO-SUM GAME THEORY
 John von Neumann and Oskar Morgenstern developed a
theory of how two-person zero-sum games should be played,
based on the following assumption:

 1) Each player chooses a strategy that enables him to do the

best he can, given that his opponent knows the strategy he is
following.

 A typical and simplest example of two-person zero-sum

game in microeconomic model is a duopoly market in which
each duopolist attempts to maximise his market share.
 Given this goal, whatever a firm gains (by increasing its
share of the market) the other firm loses (because of the
decrease in its share).

 Thus any gain of one rival is offset by the loss of the other,
and the net gain sums up to zero. Hence the name 'zero-sum
game'.
 TWO-PERSON ZERO-SUM GAME (DUOPOLISTS)…
 Assumptions
 1. The firms (players) have a given, well-defined goal.

 2. Each firm knows the strategies open to it and to its rival, or

concentrates on the most important of these strategies.

 3. Each firm knows with certainty the payoffs of all combinations of

the strategies being considered. This implies that the firm knows its
total revenue, total costs and total profit from each combination of
strategies.

 4. The actions chosen by the duopolists do not affect the total size of
the market.

 5. Each firm chooses its strategy 'expecting the worst from its rival„
(this means that each firm acts in the most conservative way),
expecting that the rival will choose the best possible counter-strategy
open to him. This behaviour is defined as 'rational'.

 6. In the zero-sum game there is no incentive for collusion, given

assumption 4, since the goals of the firms are diametrically opposed.
STRATEGIES EMPLOYED IN A TWO PERSON ZERO SUM GAME

 The strategies employed in a two person zero sum

game could be:
 i. Pure strategies

 ii. Dominating strategies

 iii. Mixed strategies

 Let‟s use these assumptions of two-person zero-sum

game (Duopolists) considered earlier to determine how
the firms should play the two-person zero-sum game in
Table 4.4
 PURE STRATEGIES-A GAME WITH A SADDLE POINT
 Maxi min aij = mini max aij is called a game with saddle point.
 i j i j
 This makes us to understand that the players in the game always use pure strategies.

 The element at the intersection of their pure strategies is known as saddle point,
which is the value of the game.

 As the players use the pure optimal strategies, the game is known as strictly
determined game.

 The saddle point is always the smallest element in the row and the greatest
element in the column.

 Not all the rectangular games will have saddle point, but if the game has saddle point,
then the pure strategies corresponding to the saddle point are the best strategies and
the number at the point of intersection of pure strategies is the value of the game.

 Once the game has the saddle point the game is solved.
 The rules for finding the saddle point are:
 1. Select the minimums of each row and encircle them.
 2. Select the maximums of each column and square them.
 3. A point where both circle and square appears in the matrix at the same point is the
saddle point.
 Another name given to saddle point is equilibrium point of the game and the
corresponding strategies form the equilibrium pair of strategies.
 TABLE 4.4: A GAME WITH A SADDLE POINT
Row Player’s Strategy Column Player’s Strategy
Column 1 Column 2 Column 3 Ro w Minimum

Row 1 4 4 10 4

Row 2 2 3 1 1

Row 3 6 5 7 5
Column Maximum 6 5 10
How should the row player play this game? If he chooses row 1, then (the
assumption implies that) the column player will choose column 1 or column 2
and hold the row player to a reward of four units (the smallest number in row 1
of the game matrix).
Similarly, if the row player chooses row 2, then (the assumption implies that) the
column player will choose column 3 and hold the row player‟s reward to one unit
(the smallest or minimum number in the second row of the game matrix).
If the row player chooses row 3, then he will be held to the smallest number in
the third row (5). Thus, the assumption implies that the row player should
choose the row having the largest minimum. Because max (4, 1, 5) is 5, the row
player should choose row 3. By choosing row 3, the row player can ensure that he
will win at least max (row minimum) of five units.
A GAME WITH A SADDLE POINT…
 From the column player‟s viewpoint, if he chooses column 1,
then the row player will choose the strategy that makes the
column player‟s losses as large as possible (and the row
player‟s winnings as large as possible).

 Thus, if the column player chooses column 1, then the row

player will choose row 3 (because the largest number in the
first column is the 6 in the third row).

 Similarly, if the column player chooses column 2, then the

row player will again choose row 3, because 5 max (4, 3, 5).

 Finally, if the column player chooses column 3, the row
player will choose row 1, causing the column player to lose
10 max (10, 1, 7) units.

 Thus, the column player can hold his losses to min (column
maximum) min (6, 5, 10), 5 by choosing column 2.
 A GAME WITH A SADDLE POINT…
 We have established that the row player can ensure that he will win at
least five units and the column player can hold the row player‟s
winnings to at most five units.

 Thus, the only rational outcome of this game is for the row player to
win exactly five units.

 He cannot expect to win more than five units because the column
player (by choosing column 2) can hold the row player‟s winnings to
five units.

 The game matrix we have just analyzed has the property of satisfying
the saddle point condition:

 max (row minimum) = min (column maximum) (1)

 all rows all columns

 Any two-person zero-sum game satisfying (1) is said to have a saddle

point.

 If a two person zero-sum game has a saddle point, then the row player
should choose any strategy (row) attaining the maximum on the left
side of (1).
 A GAME WITH A SADDLE POINT…
 The column player should choose any strategy (column) attaining the minimum
on the right side of (1).

 Thus, for the game we have just analyzed, a saddle point occurred where the row
player chose row 3 and the column player chose column 2.

 The row player could make sure of receiving a reward of at least five units (by
choosing the optimal strategy of row 3), and the column player could ensure that
the row player would receive a reward of at most five units (by choosing the
optimal strategy of column 2).

 If a game has a saddle point, then we call the common value of both sides of (1)
the value (v) of the game to the row player.

 Thus, this game has a value of 5.

 An easy way to spot a saddle point is to observe that the reward for a saddle
point must be the smallest number in its row and the largest number in its
column.

 Thus, like the center point of a horse‟s saddle, a saddle point for a two-person
zero-sum game is a local minimum in one direction (looking across the row) and a
local maximum in another direction (looking up and down the column).
 A GAME WITH A SADDLE POINT…
 A saddle point can also be thought of as an equilibrium point in that neither
player can benefit from a unilateral change in strategy.
 For example, if the row player were to change from the optimal strategy of row 3
(to either row 1 or row 2), his reward would decrease, while if the column player
changed from his optimal strategy of column 2 (to either column 1 or column 3),
the row player‟s reward (and the column player‟s losses) would increase.

 Thus, a saddle point is stable in that neither player has an incentive to move
away from it.

 Another easiest way of spotting saddle point is to follow these three steps:
 Step 1: Select the minimum (lowest) element in each row of the payoff matrix
and write them under „row minima‟ heading. Then, select the largest element
among these elements and enclose it in a rectangle.

 Step 2: Select the maximum (largest) element in each column of the payoff
matrix and write them under „column maxima‟ heading. Then select the lowest
element among these elements and enclose it in a circle.

 Step 3: Find out the element(s) that is (are the) same in the circle as well as
rectangle and mark the position of such element(s) in the matrix. This element
represents the value of the game and is called the saddle (or equilibrium) point.

 Table 4.5 presents saddle point using the three steps above. From the table,
determine the optimal strategies for players A and B. Also determine the value of
game. Is this game (i) fair? (ii) strictly determinable?
 TABLE 4.5: A GAME WITH A SADDLE (OR EQUILIBRIUM) POINT
Player B
Player A B1 B2 B3
A1 -1 2 -2
A2 6 4 -6
SOLUTION

From the table, gains to player A(losses to player B) are represented by the positive
quantities, whereas, losses to A (gains to B) are represented by negative quantities.
It is assumed that A wants to maximize his minimum gains from B.
A is concerned with the quantities of the payoffs that represent the row minimums. Now A
can do no worse than receive one of these values. The best of these values occurs when he
chooses strategy A1. This choice provides a payoff of – 2 to A when B chooses strategy B3.
This refers to A‟s choice of A1 as his maximum payoff strategy because this row contains
the maximum of A‟s minimum possible payoffs from his competitor B.
 A GAME WITH A SADDLE POINT…
 Similarly, it is assumed that B wants to minimize his losses and wishes that his
losses to A be as small as possible.
 The column maximums also represent the greatest payments B might have to
make to A.

 The smallest of these losses is – 2, which occurs when A chooses his course of
action, A1 and B chooses his course of action, B3.
 This choice of B3 by B is his minimax loss strategy because the amount of this
column is the minimum of the maximum possible losses.
 In the solution, the quantity – 2 in row A1 and column B3 is enclosed both in
the box and the circle.

 That is, it is both the minimum of the column maxima and the maximum of the
row minima. This value is the saddle point.

 The payoff amount in the saddle-point position is also called value of the game.

 For this game, value of the game is, V = – 2, for player A.

 The value of game is always expressed from the point of view of the player
whose strategies are listed in the rows.
 The game is strictly determinable.
 Also since the value of the game is not zero, the game is not fair.
 A GAME WITH A SADDLE POINT…
 Another example is a duopolictic model as presented in Table 4.6
 Choice of strategy by Firm I

 Firm I examines the outcomes of each strategy open to it. So, it will examine each
row of its payoff matrix and find the most favourable outcome of the
corresponding strategy because it expects the rival to adopt the most
advantageous action open to it. This is the behavioural rule implied by
assumption 5 of this model.

 Thus, if Firm I adopts strategy A1, the worst outcome that it may expect is a
share of 0·10 (which will be realised if the rival Firm II adopts its most favourable
strategy B1).

 If Firm I adopts strategy A2 , the worst outcome will be a share of 0-30 (if the
rival adopts the best action for him, B2).

 If Firm I adopts strategy A3 , the worst outcome will be a share of 0·20 (if Firm II
chooses the best open alternative, B3).

 If Firm I adopts strategy A4 , the worst outcome will be a share of 0·15 (which
would be realised by action B2 of Firm II).

 Among all these minima (that is, among the above worst outcomes), Firm I
chooses the maximum, the 'best of the worst„ (called a maximin strategy) because
the firm chooses the maximum among the minima.

 In Table 4.6, maximin strategy of Firm I is A2, that is, the strategy which yields a
share of 0·30.
 A GAME WITH A SADDLE POINT…
 Choice of strategy by Firm II
 Firm II behaves in exactly the same way as Firm I. The only difference is that
Firm II examines the columns of its payoff table, because these columns include
the results-payoffs of each of the strategies open to Firm II.
 For each strategy (that is, for each column), Firm II finds the worst outcome (on
the assumption that the rival will choose the best), and among these worst
outcomes Firm II chooses the best.
 Thus, if Firm II uses its own payoff table, its behaviour is a maximin behaviour
identical to the behaviour of Firm I.

 However, in the zero-sum game only one payoff matrix is adequate for the
equilibrium solution.
 So, Firm II examines the columns of the (first) payoff matrix because these
columns contain the information about the payoffs of its strategies.
 For each column-strategy, Firm II finds the maximum payoff (of Firm I) because
this is the worst situation the firm (II) will face if it adopts the strategy
correponding to that column.
 Thus, for strategy B1 the worst outcome (for Firm II) is 0·40; for strategy B2 the
worst outcome is 0·30; for strategy B3 the worst outcome is 0·50; for strategy B4
the worst result is 0·60; for strategy B5 the worst result is 0·50.
 Among these maxima of each column-strategy Firm II will choose the strategy
with minimum value. Thus the strategy of Firm II is a minimax strategy, since it
involves the choice of a minimum among the maxima payoffs. (Table 4.6.)
TABLE 4.6: PAYOFF MATRIX FOR DUOPOLISTIC FIRMS I & II
Firm II's strategies (minimax behaviour)
Firm I's strategies B1 B2 B3 B4 B5
(maximin behaviour)
A1 0·10 0·20 0·15 0·30 0·25
A2 0·40 0·30 0·50 0·55 0·45
A3 0·35 0·25 0·20 0·40 0·50
A4 0·25 0·15 0·35 0·60 0·20
Although different terms are used for the choice of the two firms (maximin
behaviour of Firm I, minimax behaviour of Firm II), the behavioural rule for
both firms is the same: each firm expects the worst from its rival.
From the Table, the equilibrium solution is strategy A2 for Firm I and B2 for
Firm II.
This solution yields shares 0·30 which is the equilibrium solution (saddle
point) because it is the preferred one by both firms while the preferred
strategies A2 and B2 are called 'dominant strategies'.
There will be no such equilibrium (saddle) solution if there is no payoff which
is preferred by both firms simultaneously.
TABLE 4.7: GAME WITH NO SADDLE POINT
Column Player’s Strategy
Row Player’s Column 1 Column 2 Row
Strategy Minimum
Row 1 -1 +1 -1
Row 2 +1 -1 -1
Column +1 +1
Maximum

It is not the case that games must have saddle point.

Many two-person zero-sum games do not have saddle
points.
For example, the game in Table 4.7 does not have a
saddle point, because as we can see max (row minimum)
= -1 < min (column maximum) = +1
 GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES
 Principle of Dominance in Games
 If there is a case of no saddle point, we can use dominating strategy to solve the
game problem.

 This involves reduction of the given matrices as the case may be.

 For instance a given game matrix (m × n) may be reduced to m × 2 or 2 × n or 2

× 2 matrix, which will help us to proceed further to solve the game.
 The ultimate way is we have to reduce the given matrix to 2 × 2 to solve
mathematically.

 The general rules of dominance can be formulated as below

 1. If all the elements of a column (say ith column) are greater than or equal to
the corresponding elements of any other column (say jth column), then ith
column is dominated by jth column.

 2. If all the elements of rth row are less than or equal to the corresponding
elements of any other row, say sth row, then rth row is dominated by sth row.

 3. A pure strategy of a player may also be dominated if it is inferior to some

convex combinations of two or more pure strategies, as a particular case, inferior
to the averages of two or more pure strategies.

 To discuss the principle of dominance, let us consider the matrix given in Table
4.8.
TABLE 4.8: GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…

Player B’s Strategies

Player A’s B1 B2 B3 B4 Row
Strategies Minimum
A1 2 -4 -3 4 -4
A2 4 -3 -4 2 -4
Column 4 -3 -3 4
Maximum
The row minimums and column maximums show that the problem is not having saddle
point.
Hence we have to use method of dominance to reduce the size of the matrix.
(i) Starting with B‟s strategies, if B plays the first strategy, he looses 2 units of money
when A plays first strategy and 4 units of money when A plays second strategy.
Similarly, considering B‟s second strategy, B gains 4 units of money when A plays his
first strategy and gains 3 units of money when A plays second strategy.
Irrespective of A‟s choice, B will gain money. Hence for B his second strategy is
superior to his first strategy. In other words, B's second strategy dominates
B's first strategy. Or B’s first strategy is dominated by B's second strategy.
Hence we can remove the first strategy of B from the game to have a reduced matrix as
shown in Table 4.9.
 TABLE 4.9: GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…

Player B’s Strategies

Player A’s B2 B3 B4
Strategies
A1 -4 -3 4
A2 -3 -4 2
(ii) Consider B‟s B3 and B4 strategies. When B plays B4 strategy, he
loose 4 units of money when A plays his first strategy and 2 units of
money when A plays his second strategy.

Whereas, when B plays his B3 strategy, he gains 3 units of money

and 4 units of money, when A plays his A1 and A2 strategy
respectively.

Hence B‟s B4 strategy (pure strategy) is dominating the third

strategy.
Therefore, we can remove the same from the game to have the
reduced matrix presented in Table 4.10.
 TABLE 4.10: GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…

Player B’s Strategies

Player A’s B2 B3
Strategies
A1 -4 -3
A2 -3 -4
Looking at Table 4.8, it is clear that the elements of second column are smaller
or less than the elements of column 4.
Similarly elements of 3 column also smaller or less than the elements
of columns1 and 4.
The implication is that the dominance rule for columns means that
when elements of a column, say ith are less than or equal to the
corresponding elements of jth column, then jth column is
dominated by ith column or ith column dominates jth column.
For that of row, consider the matrix given in Table 4.11.
 TABLE 4.11: GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…
Player B’s Strategies
Player A’s Strategies B1 B2
A1 -2 -4
A2 1 2
From the Table, if A plays his first strategy, then he looses 2 units of money
and looses 4 units of money when B plays his second strategy.
But when A plays his second strategy, he gains 1 unit of money for B‟s first
strategy and gains 2 units of money, for B‟s second strategy.
Hence, A's second strategy (pure strategy) is superior to A's first
strategy or A's second strategy dominates A's first strategy or A's
first strategy is dominated by A's second strategy.
From close examination, it is evident that elements of A's second
strategy are greater than the elements of first strategy.
Therefore. The general rule of dominance for rows is that when the
elements of rth row are greater than or equal to elements of sth
row, then rth row dominates sth row or sth row is dominated by rth
row.
GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…

 Note that for Dominating or dominant strategies, at

every reduction of the matrix, check for the existence
of saddle point.

 If saddle point is found, the game is solved.

 Otherwise continue to reduce the matrix by method of

dominance until a saddle point is discovered.
 GAMES WITH NO SADDLE- MIXED STRATEGIES
 In rectangular games, when we have saddle point, the best strategy
is pure strategy.

 When games do not have saddle points and cannot be solved by

dominating strategy, the best strategy to use is a mixed strategy.

 In this case what we do is to randomise.

 Random choice of strategies is the main idea behind a mixed

strategy.

 Generally a mixed strategy for player is defined as a proba6iffty

distribution on the set of pure strategies.

 The minimax theorem put forward by von Neumann enables us to

find the optimal strategies and value of a game that has no saddle
point solution and he was able to show that every two-person zero-
sum game has a solution in mixed if not in pure strategy.

 So. while dealing with mixed strategies, we have to determine the

probabilities with which each action should be selected.
 GAMES WITH NO SADDLE- MIXED STRATEGIES…
 Whenever a game does not possess a saddle point, game theory advises each
player to assign a probability distribution over her set of strategies.

 To express this mathematically, let

 xi = probability that player 1 will use strategy i (i 1, 2, . . . , m),
 yj = probability that player 2 will use strategy j ( j 1, 2, . . . , n),
 where m and n are the respective numbers of available strategies.
 Thus, player 1 would specify her plan for playing the game by assigning values
to x1, x2, . . . , xm.

 Because these values are probabilities, they would need to be nonnegative and
add to 1.

 Similarly, the plan for player 2 would be described by the values she assigns to
her decision variables y1, y2, . . . , yn.

 These plans (x1, x2, . . . , xm) and (y1, y2, . . . , yn) are usually referred to as
mixed strategies, and the original strategies are then called pure strategies.

 When the game is actually played, it is necessary for each player to use one of
her pure strategies.

 However, this pure strategy would be chosen by using some random device to
obtain a random observation from the probability distribution specified by the
mixed strategy, where this observation would indicate which particular pure
strategy to use.
GAMES WITH NO SADDLE- MIXED STRATEGIES…
 When probabilities are assigned, the game can the be
played and analysed using the expected payoff.

 So the expected payoff for player 1 will be computed

using
m n
E ij   p ij xi y j
i 1 i 1

 where pij is the payoff if player 1 uses pure strategy i

and player 2 uses pure strategy j.

 If probability is not given, we can compute it

 TABLE 4.12: GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…

Player B’s Strategies and probabilities

y1 y2
Player A’s Strategies and probabilities B1 B2
x1 A1 a11 a12
x2 A2 a21 a22
From the 2 × 2 game given in the Table, Let x1 and x2 be the probability with
which A plays his first and second strategies respectively.
Similarly B plays his first and second strategies with probability of y1 and y2
respectively.
Now x1 + x2 = 1, and y1 + y2 = 1.
Let us work out expected gains of A and B when they play the game with
probabilities of x1, x2 and y1and y2.
A’s expected gains when:
B plays his first strategy = a11 x1 + a21 x2
When B plays his second strategy = a12 x1 + a22 x2
Similarly B's gains when:
A plays his first strategy = a11 y1 + a12 y2
When A plays his second strategy = a21 y1 + a22 y2
 GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…

 Assume that v is the value of the game.

 As A is the maximin player, he wants to see that his gains are ≥ v.
 As B is the minimax player, he wants to see that his gains must be
always ≤ v.
 Therefore, we have:
 a11 x1 + a21 x2 ≥ v
 a12 x1 + a22 x2 ≥ v and
 a11 y1 + a12 y2 ≤ v
 a21 y1 + a22 y2 ≤ v

 To find the value of x1, x2 and y1, y2 we have to solve the above
given inequalities.
 For convenience, let us consider them as equations to find the
values of x1, x2 and y1, y2. Therefore, we have:
 a11 x1 + a21 x2 = v
 a12 x1 + a22 x2 = v and
 a11 y1 + a12 y2 = v
 a21 y1 + a22 y2 = v
GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…
 Always we workout a solution of a 2 × 2 game by considering
the above inequalities as strict equalities.
 Now we can write the above as:
 a11x1 + a21x2 = v = a12x1 + a22x2
 This can be written as x1(a 11 – a 12 ) = x2( a 22 – a 21 ) or
 (x1/x2 ) = (a22 – a21)/(a11 – a12)

 Similarly we can write:

 (y1/y2) = (a22 – a12)/(a11 – a12 ), by simplifying, we get:
 x1 = ( a22 – a21 ) / ( a 11 + a22) – (a12 + a21) or = 1 – x2
 x2 = (a11 – a12) / ( a11 + a22) – (a12 + a21) or = 1 – x1
 y1 = (a22 – a12) / (a11 + a22) – (a12 + a21) or = 1 – y2
 y2 = (a 11 – a 21) / (a11 + a22) – (a12 + a21) or = 1 – y1,
 and the value of the game is
 v = (a11 a22 - a 12 a21) / ( a11 + a22) – (a12 + a21)
 GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…
 Note that:
 For x1 Numerator = a22 – a21 i.e. x1 is in the first row, for numerator we have to
take the difference of second row elements from right to left.
 For x2, which comes in second row, we have to take difference of the first row
elements from left to right.

 For y1 which comes in the first column, we have to take the difference of second
column elements from bottom to top.

 For y2, which comes in second column, we have to take the difference of the
elements of first column from top to bottom.

 As for the denominator is concerned, it is common for all formulae. It is given by

sum of diagonal elements from right hand top corner to left-hand bottom corner
minus the sum of the elements diagonally from left-hand top corner to right hand
bottom corner.

 For value of the game, the numerator is given by products of the elements in
denominator in the first bracket minus the product of the elements in the second
bracket.

 When the game does not have saddle point, the two largest elements of its payoff
matrix must constitute one of the diagonals.
 Using the information in Table 4.10, which was arrived at after reducing Table
4.8
 GAMES WITH NO SADDLE POINT-DOMINATING STRATEGIES…
Player B’s Strategies
Player A’s ( y1) B2 (y2) B3 Row Minimum
Strategies
(x1) A1 -4 -3 -4
(x2) A2 -3 -4 -4
Column -3 -3
Maximum
From the Table, our probabilities and the value of the game are computed as:
x1 = (a22 – a21) / (a11 + a22) – (a12 + a21) or = 1 – x2
x1 = (–4 – [–3]) / (–4 + [–4]) – (–3 + [–3]) = (–4 + 3) / ( –4 – 4) – (–3 –3) = –1 / (–
8) – (–6) = –1 / –8 + 6 = –1 / –2 = 1/2 = 0.5.
x2 = 1 – x1 = 1 – 0.5 = 0.5.
y1 = (a22 – a12) / (a11 + a22) – (a12 + a21) or = 1 – y2 = [–4 + (–3)] / [–4 + ( –3)]
– [–3 + (–3)] = (–4 + 3) / (–4 – 3) – (–3 –3) = –1 / (–7 + 6) = 1 (i.e. pure strategy).
Value of the game = v = (a11 a22 – a12 a 21) / (a11 + a22) – (a12 + a21)
[12 – 12) / [`–4 – 3] – [–3 – 3] = 0
 PRACTICE QUESTIONS
QUESTION 1 Player B’s Strategies

Player A’s Strategies B1 B2 B3

A1 1 7 2
A2 6 2 7
A3 5 1 6
Solve the game using the payoff matrix
QUESTION 2
Player A’s Strategies B1 B2 B3 B4
A1 3 2 4 0
A2 3 4 2 4
A3 4 2 4 0
A4 0 4 0 8
Use the concept of dominance to solve the game in question 2.
 PRACTICE QUESTIONS . . .
 Question 3: Two competing telecommunication companies MTN and Airtel both
have objective of maintaining large share in the telecommunication industry.
 They wish to take a decision concerning investment in a new promotional
campaign.
 Airtel wishes to consider the following options:
 r1: advertise on the Internet
 r2: advertise in all mass media
 MTN wishes to consider these alternatives
 c1: advertise in newspapers only
 c2: run a big promo

 If Airtel advertises on the Internet and MTN advertises in newspapers, MTN will
increase its market share by 3% at the expense of Airtel.
 If MTN runs a big promo and Airtel advertises on the Internet, Airtel will lose
2% of the market share.
 If Airtel advertises in mass media only and MTN advertises in newspapers,
Airtel will lose 4%.
 However, if Airtel advertises in mass media only and MTN runs a big promo,
Airtel will gain 5% of the market share.
 a) present this information on a payoff table
 b) What is the best policy that each of the two companies should take?
 GAME THEORY USING DECISION TREE
 Basic notions
 There are two ways of representing a game: the extensive form and the
strategic form.

 Let‟s begin by analyzing the main elements of an extensive form game.

 Games in extensive form (dynamic or sequential games)
 An extensive form game specifies:

 1) The players.

 2) The order of the game.

 3) The choices available to each player at each turn of play (at each decision
node).

 4) The information held by each player at each turn of play (at each decision
node).

 5) The payoffs of each player as a function of the movements selected.

 6) Probability distributions for movements made by nature.

 GAME THEORY USING DECISION TREE…
 An extensive form game is represented by a decision tree.
 A decision tree comprises nodes and branches.
 There are two types of node: decision nodes and terminal nodes.
 We have to assign each decision node to one player.
 When the decision node of a player is reached, the player chooses a
move.
 When a terminal node is reached, the players obtain payoffs: an
assignment of payoffs for each player.

 EXAMPLE 1: Entry game

 Consider a market where there are two firms: an incumbent firm,
A, and a potential entrant, B.
 At the first stage, the potential entrant decides whether or not to
enter the market.
 If it decides “not to enter” the game concludes and the players
obtain payoffs (firm A obtains the monopoly profits) and if it decides
“to enter” then the incumbent firm, A, has to decide whether to
accommodate entry (that is, to share the market with the entrant)
or to start a mutually injurious price war.
 GAME THEORY USING DECISION TREE…
 The extensive form game can be represented as follows:

 Players: B and A.
 Actions: E (to enter), NE (not to enter), Ac. (to accommodate), P.W. (price
war).
 Decision nodes: α.
 Terminal nodes: β.
 (x, y): vector of payoffs. x: payoff of player B; y: payoff of player A.
 At each terminal node we have to specify the payoffs of each player (even
though some of them have not actually managed to play).
 GAME THEORY USING DECISION TREE…
 Assumptions:
 (i) All players have the same perception of how the game
is like.

 (ii) Complete information: each player knows the

characteristics of the other players: preferences and
strategy spaces.

 (iii) Perfect recall (perfect memory): each player

remembers her previous behavior in the game.

 Definition 1: Information set

 “The information available to each player at each one of
her decision nodes”.
 GAME THEORY USING DECISION TREE…

 In game 1, player 2 has different information at each one of her decision nodes. At node A, if
she is called upon to play she knows that player 1 has played I and at B she knows that
player 1 has played D.

 We say that these information sets are singleton sets consisting of only one decision node.
 Perfect information game: a game where all the information sets are singleton sets or, in
other words, a game where all the players know everything that has happened previously in
the game.

 In game 2, the player 2 has the same information at both her decision nodes. That is, the
information set is composed of two decision nodes. Put differently, player 2 does not know
which of those nodes she is at.

 A game in which there are information sets with two or more decision nodes is called an
imperfect information game: at least one player does not observe the behaviour of the
other(s) at one or more of her decision nodes.

 The fact that players know the game that they are playing and the perfect recall assumption
restrict the situations where we can find information sets with two or more nodes.
 GAME THEORY USING DECISION TREE…

 Game 3 is poorly represented because it would not be an imperfect

information game.

 Assuming that player 2 knows the game, if she is called on to move and
faces three alternatives he/she would immediately deduce that the
player 1 has played I.

 That is, the game should be represented like game 4.

 Therefore, if an information set consists of two or more nodes the
number of alternatives, actions or moves at each one should be the
same.
 GAME THEORY USING DECISION TREE…
 Definition 2: Subgame
 “It is what remains to be played from a decision node with the condition that
what remains to be played does not form part of an information set with two or
more decision nodes.
 To build subgames we look at parts of the game tree that can be constructed
without breaking any information sets.
 A subgame starts at a singleton information set and all the decision nodes of
the same information set must belong to the same subgame.”
 EXAMPLE 2: The Prisoner‟s Dilemma
 Two prisoners, 1 and 2, are being held by the police in separate cells.
 The police know that the two (together) committed a crime but lack sufficient
evidence to convict them.

 So the police offer each of them separately the following deal: each is asked to
implicate his partner.
 Each prisoner can “confess” (C) or “not confess” (NC).
 If neither confesses then each player goes to jail for two months.
 If both players confess each prisoner goes to jail for one month.

 If one prisoner confesses and the other does not confess, the first player goes
free while the second goes to jail for three months.
 GAME THEORY USING DECISION TREE…
 Each suspect has two 'strategies' open to himself, to confess or not to
confess.

 He faces is faced with the dilemma: to confess (and go free if the other does not
confess while the second goes to jail for six months.

 Or get two months jail term if neither confesses. Or get a month jail if both confess.
Or confess and goes free if the other prisoner did not confess.

 Simultaneous case: each player takes her decision with no knowledge of the decision
of the other.

 This is an information set with two decision nodes. This is an imperfect information
game. There is a subgame which coincides with the proper game.
 REFERENCES
 Adebayo, O.A. et al. (2006). Operations Research in Decision and Production
Management.
 Aguirre, A. (2023). An Introduction to Game Theory: Market Power and Strategy.
Department of Economic Analysis, University of the Basque Country UPV/EHU.
Available at
https://www.ehu.eus/iaguirre/An%20Introduction%20to%20Game%20Theory.pdf
 Babatunde, S.J. & Hassan, S.Q.O. (2015). Operations Research. NOUN.
 Dixon – Ogbechi, B.N. (2001). Decision Theory in Business. Lagos: Philglad Nig. Ltd.
 Gupta, P.K. & Hira, D.S. (2012). Operations Research, New Delhi: S. Chand &
Company.
 Hillier, F.S & Lieberman, G.J. (2010). Introduction to Operations Research (Ninth
Edition). Published by McGraw-Hill.
 Koutsoyiannis, A. (2003). Modern Microeconomics. (2nded.). Macmillan Press Ltd.
 Murthy, P. R. (2007). Operations Research (Second Edition). New Age International
(P) Ltd., Publishers.
 scikit-learn.org. (2023). Decision Trees https://scikit-
learn.org/stable/modules/tree.html.
 Sharma, J.K. (2017). Operations Research: Theory and Applications (Sixth Edition).
Trinity Press.
 Taha, H.A. (2017). Operations Research: An Introduction (Tenth Edition). Pearson
Education Limited.
 Taylor III, B.W. (2007). Introduction to Management Science. (9th ed.). New Jersey:
Pearson Education Inc.
 Winston, W.L & Goldberg, J.B. (2004). Operations Research: Applications and
Algorithms. Brooks/Cole, a division of Thomson.