CHP - 48 - Competitive Strategies (Games Theory)
CHP - 48 - Competitive Strategies (Games Theory)
CHP - 48 - Competitive Strategies (Games Theory)
Chapter
Competitive
Strategies
(Games Theory)
Chapter Outcomes
After completing the chapter, you will be able to:
OO Define a competitive game.
OO Describe characteristics of a competitive game.
OO Define two person Zero sum Game.
OO Apply maximin and minimax principle to solve the game by Saddle point.
OO Use algebraic method to solve 2 × 2 game.
OO Reduce the size of the game by applying dominance properties.
OO Apply graphical method to solve the game.
OO Formulate games theory problem as LPP problem and solve the problem as LPP.
48.1 INTRODUCTION
Von Nuemann developed the theory of games in 1928 and he is considered as father of
Games theory. Games theory is a type of decision theory in which one’s choice is determined
after considering possible alternatives available to the opponent playing the game. In
ordinary decisions under uncertainty, the decision maker is faced with only a random
process. The decision maker has not only to analyse alternative course of action available
to him but also consider the possible goals, strategies and choices of the competitor.
Thus, many practical problems require decision making in a competitive situation
where there are two or more opposing parties with conflicting interests and where the
action depends upon the one taken by the opponent e.g. candidates contesting for an
election, advertising and promotional campaigns by competing business firms etc. In a
competitive situation, with a finite number of competitors is called “a competitive game”.
The games where there are number of competitive actions (or choices) usually referred to
1123
1124 Industrial Engineering and Production Management
as games of strategy because a person playing one of the these games is merely betting
against the odds.
Player B
B1 B2 B3 Row minimum
A1 9 2 3 2
Player A A2 8 6 7 6
Saddle point
A3 6 4 5 4
Column maximum 9 6 7
Suppose the players adopt an over cautious approach, and if they do so, then, they will
assume the worst and act accordingly. Suppose if player A adopts a pessimistic approach
to the pay-off matrix, then the worst outcome of A for his strategy A1 will be 2. (A player
A’s minimum gain).
Similarly, the minimum outcome for player A’s strategy A2 and A3 will be 6 and 4
respectively.
A is a maximising player and he aims at maximising his minimum gains. Thus, the
maximum of all the minimum values is 6 . So the best strategy for A is A2 which maxim-
ises his minimum gains. Thus, maximin refers to maximising minimum gains. Player B is
a minimising player and he tries to minimise the maximum gains of player A.
The maximum pay-off of the player A corresponding to the B’s strategies B1, B2 and B3
are [9, 6, 7]. Thus, B wants to minimise these maximum gains. The minimum of maximum
value of pay-off is 6 . So, player ‘B’ always tries to use his strategy B2.
Thus, minimax refers to minimising maximum gains of A.
Saddle Point
Saddle point in a pay-off matrix is the position in the matrix where the maximum of
row minimum (maximin) coincides with the minimum of column maximum (minimax).
The pay-off at the saddle point is called the value of the game which equals minimax of
miximin value.
The optimum solution can be had for the game by applying minimax and maximin.
Thus, the solution of the game is:
Best strategy for A is A2 ; Best strategy for B is B2; Value of the game is 6.
Steps to find out saddle point
1. Find out the minimum element of each row of the pay-off matrix and circle the
maximum value.
2. Find out the maximum element in each column of the pay-off matrix and circle the
maximum value [].
3. If these two values are same (i.e. the pay-off containing both ‘’ and marks, is
the saddle point of the matrix.
Problem: Solve the game where the pay-off matrix is given by
Player B
B1 B2 B3
Player A A1 1 3 1
A2 0 – 4 –3
A3 1 5 –1
1126 Industrial Engineering and Production Management
Solution: Mark row minimum and column maximum for each row and column.
Player B
Row minimum
B1 B2 B3
Player A A1 1 1
3
A2 0 – 4 –3
A3 1 5 –1
Column Maximum 1 5 1
A1 –2 0 0 5 3 –2
Player A A2 4 2 1 2 5 1
Maximin value
A3 – 4 –3 0 –3 6 – 4
A4 5 1 –5 2 – 6 6
Max value
5 2 1 5 6
Let x1 and x2 be the probabilities of selecting the strategies A1 and A2 by player ‘A’.
(here x2 = (1 – x1) and y1 and y2 be the probabilities of selecting strategies B1 & B2 by player
B (y2 = (1 – y1).
The optimal strategies for the players are:
a22 − a21
For the player A, x1 =
a11 + a22 − a21 − a12
a11 − a12
x2 = (1 – x1) =
a11 + a22 − a21 − a12
a22 − a12
For the player B, y1 =
a11 + a22 − a21 − a12
a11 − a21
y2 =
a11 + a22 − a21 − a12
a11 a22 − a12 a21
Value of the game (V) =
a11 + a22 − a21 − a12
(Note that the denominator in all the formulas are the same).
Problem: For the pay-off matrix given below, find the optimal strategies for the players and
value of the game.
Player B
B1 B2
A1 8 –3 (x1)
Player A
A2 –3 1 (x2)
(y1) (y2)
Solution: In the above problem, saddle point does not exist. So, it is mixed strategy
problem. Let x1 and x2 represent the probabilities with which player A will use his strategies
A1 and A2 and y1 and y2 be the probabilities with which the player B chooses his strategy
B1 and B2. The optimum mixed strategies are:
1 − (−3) 4 11
x1 = = ; x2 = (1 – x1) =
8 + 1 − (−3 − 3) 15 15
For the player ‘B’, the probability with which he uses strategies B1 and B2 are:
1128 Industrial Engineering and Production Management
1 − (−3) 4 11
y1 = = ; y2 = (1 – y1) =
8 + 1 − (−3 − 3) 15 15
The expected value of the game is
8 − (−3)(−3) 1
V = =−
8 + 1 − (−3 − 3) 15
48.7 DOMINANCE PROPERTY
Principle of dominance is applicable to both pure strategies and mixed strategies.
Sometimes, it is observed that one of the pure strategies of either players is always inferior
to at least one of the remaining strategies. The superior strategies are said to dominate
the inferior ones. The player would have no incentive to use inferior strategies which are
dominated by the superior ones. In such cases of dominance, the size of the pay-off matrix
by deleting those strategies which are dominated by the others.
The dominance properties are:
1. If all the elements of a row say Kth, are less than or equal to the corresponding
elements of any other row (say rth row), then kth row is dominated by the rth row.
2. If all the elements of column, say kth, are greater than or equal to the corresponding
elements of any other column, say rth then Kth column is dominated by the rth row.
3. A pure strategy may be dominated, if it is inferior to average of two or more other
pure strategies.
Problem: The pay-off matrix of the player ‘A’ is given.
Player B
I II III IV V
I 2 4 8 3 4
II 5 6 7 3 8
Player A
III 6 7 8 9 7
IV 4 2 4 8 3
Using dominance property obtain the optimum strategies for both the players and
determine the value of the game.
Solution: As all the elements of column I are either less or equal (≤) to respective
elements of column III, column I is said to have dominance over column III and column III
can be eliminated. Hence, we get the pay-off matrix as shown.
Player B
I II IV V
I 2 4 3 4
II 5 6 3 8
Player A
III 6 7 9 7
IV 4 2 8 3
As all the elements of row III are greater than the respective elements of row IV, row
III is said to have dominance over row IV, and hence, the IV can be eliminated. Hence,
we get
Competitive Strategies (Games Theory) 1129
Player B
I II IV V
I 2 4 3 4
Player A II 5 6 3 8
III 6 7 9 7
As all the elements of column I are less than the respective elements of column V,
column I is said to dominate column V and hence, column V is eliminated.
Player B
I II IV
I 2 4 3
Player A II 5 6 3
III 6 7 9
All the elements of row III are greater than the respective elements of row I and II and
now row III is said to have dominance over row I and row II and hence, eliminating row I
and row II, we get
Player B
Player A I II III
III 6 7 9
Now, the minimum value of row III is 6 and hence it is the saddle point
Hence, the best strategy for A is III; the best strategy for B is I; and value of the game
is 6.
Problem: Solve the following game
Player B
1 2 3 4
1 3 2 4 0
2 3 4 2 4
Player A
3 4 2 4 0
4 0 4 0 8
Solution: Since all the elements of 3rd row are greater than or equal to the correspond-
ing elements of 1st row, third (3rd) row is dominating 1st row and hence row 1 can be elimi-
nated.
Player B
1 2 3 4
1 3 4 2 4
Player A
3 4 2 4 0
4 0 4 0 8
1130 Industrial Engineering and Production Management
Again, all the elements of the 1st column are greater than or equal to the corresponding
elements of the 3rd column.
∴ 3rd column is dominating the first column. Eliminating 1st column, the matrix is
reduced to
Player B
2 3 4
2 4 2 4
Player A
3 2 4 0
4 4 0 8
Here, the linear combination of 2nd and 4th column dominates the 2nd, because
2+4 4+0 0+8
4 > , 2= and 4 =
2 2 2
Eliminating 1st column, the reduced matrix becomes,
Player B
3 4
2 2 4
Player A
3 4 0
4 0 8
Here again, the convex linear combination of 3 and 4 of player ‘A’ dominate 2nd
because
4+0 0+8
2 = , 4=
2 2
∴ Eliminating 1st row, the reduced matrix becomes
Player B
3 4
3 4 0
Player A
4 0 8
2 1
∴ y2 = (1 – y1) = 1 − =
3 3
Thus, optimum strategies are:
For player A, [0, 0, 2/3, 1/3]; For player B, [0, 0, 2/3, 1/3]
and the value of the game is
a11 a22 − a21 a12 4×8−0×0 32 8
= = =
a11 + a22 − ( a12 + a21 ) 4 + 8 − (0 × 0) 12 3
Problem: Solve the following game by using the principle of dominance
Player B
B1 B2 B3 B4 B5 B6
A1 4 2 0 2 1 1
A2 4 3 1 3 2 2
Player A A3 4 3 7 –5 1 2
A4 4 3 4 –1 2 2
A5 4 3 3 –2 2 2
Now, none of the single row or column dominates another row or column i.e. none of
the pure strategies of A and B is inferior to any of the other strategies.
However, column B5 is dominated by the average of column B3 and B4.
1 + 3 7 − 5 4 − 1
, , or [2, 1, 3/2]
2 2 2
This strategy of B is superior to strategy B5 because the B5 strategy will result him in
greater loss. So, the strategy B5 can be eliminated. The modified matrix is
1132 Industrial Engineering and Production Management
Player B
B3 B4
A2 1 3
Player A A3 7 –5
A4 4 –1
1+7 3 − 5
The average of A’s pure strategies A2 and A3 , or (4, –1) which is obvi-
2 2
ously same as strategy A4. Thus, strategy A4 can be eliminated.
Thus we get the resulting matrix (2 × 2) is given by
Player B
B3 B4
A2 1 3
Player A A3 7 –5
This diagram reveals that B will never play y3in which case his loss will be more than
what it is by playing with strategy y2 as V + 3x1 ≤ 6 (representing strategy y3) which is above
V + 3x1 ≤ 5 (representing strategy y2). Eliminating this strategy for player B, we can plot the
graphs for player B.
Equations (2), (6) and (7) can be rewritten as
y1 + y2 = 1 ...(11)
4y1 + 2y2 ≤ V ...(12)
2y1 + 5y2 ≤ V ...(13)
∴ y2 = 1 – y1 Substituting for y2 in the equations (12) becomes
4y1 + 2 (1 – y1) ≤ v i.e. 2 y1 + 2 ≤ V
i.e. V – 2y1 ≥ 2 ...(14)
Equation (13) becomes
2y1 + 5 (1 – y1) ≤ V, i.e. – 3y1 + 5 ≤ V
i.e. V + 3y1 ≥ 5 ...(15)
From equation (14), when y1 = 0, V = 2
y1 = 1, V = 4
From equation (15), when y1 = 0, V = 5
y1 = 1, V = 2
Plotting these graphs, we have feasible region above CEB as shown in Fig. 48.2.
5 5
V+3y1>5
4 B
4
31/5
3 E 3
2 V+2y1>5 2
D
A
1 1
0 3/5 I
y1
Fig. 48.2: Plot of y1 Versus V
B’s objective is to minimise the value of V, this arises in the feasible region at point E.
Giving the values as:
1 3 2
V = 3 , y1 = , y2 = 1 – y1 = ,
5 5 5
48.9 SOLUTION OF GAME BY LINEAR PROGRAMMING APPROACH
Game theory has a strong relationship with linear programming since every finite two
person zero sum game can be expressed as a linear programme and conversely every
linear programme can be represented as a game.
The linear programming approach for the game can be represented as follows:
Let player ‘A’ has ‘m’ course of action (A1, A2, A3 ..... Am) and player ‘B’ has ‘n’ courses
of action (B1, B2, B3 ......Bn) The pay-off to player ‘A’ if he selects strategy Ai and player ‘B’
selects strategy Bj is aij.
Competitive Strategies (Games Theory) 1135
Mixed strategy for player ‘A’ is defined by the probabilities p1, p2, p3 ..... pm
where p1 + p2 + pm = 1.
and mixed strategy for player ‘B’ is defined by the probabilities q1, q2, ...... qn where
q1 + q2 + q3 + ..... + qn = 1 let ‘V’ be value of game to A. Then, the game can be defined as a
linear programming problems as follows:
Determine the unknown quantities p1, p2, ..... pn with an objective is to maximise the
value of the game ‘V’ such that the following constraints are satisfied.
a11p1 + a21p2 + ...... + am1pm ≥ V
a12p1 + a22p2 + ..... + am2pm ≥ V
: :
: :
a1np1 + a2np2 + ....... + amnpm ≥ V
p1 + p2 + p3 + ....pm = 1
p1, p2, p3 .... pm ≥ 0
If the pay-off’s i.e. aij are negative, which is not allowed in linear programming sim-
plex method. To prevent this, a constant ‘K’ can be added to all the elements of the pay-off
matrix to make them all non-negative. Once the optimal solution is obtained for L.P.P., the
constant ‘k’ can be subtracted from the objective value to get the true value of the game.
Now, dividing each inequality by ‘V’, we get
a11 (p1/V) + a21 (p2/V) + ... + am1 (pm/V) ≥ 1
a12 (p1/V) + a22 (p2/V) + ... + am2 (pm/V) ≥ 1
:
:
a1n (p1/V) + a2n (p2/V) + ... + amn (pm/V) ≥ 1
(p1/V) + (p2/V) + p3/V + ... + pm/V ≥ 1/V
To simplify the problem, we define new variables.
Let p1/V = x1 , p2/V = x2 , ..... pn/V = xn
The objective of the firm is to maximise the value of V, which is equivalent to minimising
1/V.
The resulting linear programming problem can be written as,
1
Minimize Z = or Z = x1 + x2 + ... + xn
V
Subjected to constraints,
a11x1 + a21x2 + ... + am1xn ≥ 1
a12x2 + a22x2 + ... + am2xn ≥ 1
:
:
a1nx1 + an2x2 + ... + amnxn ≥ 1
x1, x2 ..... xn ≥ 0
Example: Consider payoff matrix with respect of Player A and solve it optimally:
B
1 2
1 6 9
A
2 8 4
Solution: The maximum and minimax values of the given problem are shown in
below:
B
1 2 Maximin
1 6 9 6 (maximum)
A
2 8 4 4
8 9
(minimax)
In this problem, the maximin value (6) is not equal to the minimax value (8). Hence,
the game has no saddle point. Under this situation, the formulae given in this section are
to be used to find the mixed strategies of the players and also the value of the game. The
computations of oddments of the game are sumarized in Table below:
Table 48.3: Payoff matrix with Oddments
B
1 2 Oddments
1 6 9 4
A
2 8 4 3
Oddments 5 2
Let p1 and p2 be the probabilities of selection Alternative 1 and of Alternative 2,
respectively of Player A. Also q1 and q2 be the probability of selection of Alternative 1 and
of Alternative 2 of Player B, respectively. Then, we have
| c − d| 4 4
p1 = = =
| a − b| +|c − d| 3 + 4 7
| a − b| 3 3
p2 = = =
| a − b| +|c − d| 3 + 4 7
| b − d| 5 5
q1 = = =
| a − c| + | b − d | 2 + 4 7
| a − c| 2 2
q2 = = =
| a − c| + | b − d | 2 + 5 7
where, the value of the game is
a|c − d| + c| a − b| (6)(4) + (8)(3) 48
V = = =
| a − b| +|c − d| 4+3 7
Hence, the strategies of Player A is: A(4/7, 3/7) and of Player B: B(5/7, 2/7)
48 6
The value of the game = =6
7 7
Competitive Strategies (Games Theory) 1139
SUMMARY
Games or ‘Strategic Interactions’ can be found in all walks of life. Examples of such scenarios
are two firms competing for market share, politicians contesting elections, different
bidders participating in an auction for wireless spectrum, coal blocks etc. Game theory
provides a convenient framework to model and interpret the behavior of participants in
such strategic interactions. Hence it can be applied to solve a wide variety of problems
involving diverse areas such as Markets, Auctions, Online Retail, Cold War, Paying Taxes,
Bargaining, Elections, Portfolio Management etc.
Developed in 1950s by mathematicians John von Neumann and economist Oskar
orgenstern. Games Theory Designed to evaluate situations where individuals and
organizations can have conflicting objectives. A game is said to be zero-sum if wealth is
neither created nor destroyed among the players. A game is said to be non-zero-sum if
wealth may be created or destroyed among the players (i.e. the total wealth can increase
or decrease).
Game theory is the study of how people behave in strategic situations. Strategic
decisions are those in which each person, in deciding what actions to take, must consider
how others might respond to that action. Four elements to describe a game:
• Players: Participants
• rules: when each player moves, what are the possible moves, what is known to
each player before moving;
• outcomes of the moves;
• payoffs of each possible outcome: how much money each player receive for any
specific outcome.
A strategy is a plan of action by which a player has a decision rule to determine their
set of moves for every possible situation in a game. A strategy is said to be pure if it at every
stage in the game it specifies a particular move with complete certainty. A strategy is said
to be mixed if it applies some randomisation to at least one of the moves.
For each game, there are typically multiple pure strategies.
Dominant strategy: One firm’s best strategy may not depend on the choice made by
the other participants in the game. Leads to Nash equilibrium because the player will use
the dominant strategy and the other will respond with its best alternative.
Dominated strategies: An alternative that yields a lower payoff than some other
strategies a strategy is dominated if it is always better to play some other strategy, regardless
of what opponents may do. It simplifies the game because they are options available to
players which may be safely discarded as a result of being strictly inferior to other options.
There are many solution methods to solve the games
1. Saddle point method
2. Games without saddle point Algebraic method, Oddment method
3. Dominance properties to reduce size of the game
4. Graphical method
5. Solution of games using LPP
REVIEW QUESTIONS
1. Explain the following terms with reference to Games theory
(i) Competitive game
(ii) Two person Zero sum game
(iii) Pay off and Pay off matrix
(iv) Saddle point and Value of the game
(v) Pure and Mixed strategies
(vi) Dominance Property
2. Illustrate with an example how to use dominance properties to reduce the size of rectangular
game?
3. Formulate games theory problem as an LPP problem
4. What are the limitations of games theory?
5. What are the solution methods available for games without saddle points?
PROBLEMS
1. Two companies manufacturing air conditioners compete for supplying slit AC’s to a new
five star hotel to be constructed. Each company has listed its alternatives/strategies
for selling AC’s. The strategies of Company A are as follows; (i) giving special discount
(ii) Giving additional warranty (iii) Special gifts. Company B the strategies are (i) Special price
offer (ii) 20% additional discount on other products (iii) Fre training to users. The estimated
gains and losses in lakhs of rupees for company A for various combination of alternatives
of both companies are summarized in the table below. Find the optimum strategies for or
companies?
B
1 2 3
I 40 45 50
A
II 20 45 60
III 25 30 30
2. Solve the following Game using dominance rules
Player B
1 2 3 4 5
1 6 15 30 21 6
Player A
2 3 3 6 6 4
3 12 12 24 36 3
Competitive Strategies (Games Theory) 1141
1 2 3 4 5
A 1 4 2 5 –6 6
2 7 –9 7 4 8
4. Consider the payoff matrix for player A and determine optimum strateguies using Graphical
method.
1 2
1 5 3
A 2 6 4
3 2 –7
4 9 –8
5. The labor union and management of a particular company have been negotiating a new labor
contract. However, negotiations have now come to an impasse, with management making
a “final” offer of a wage increase of 80 per hour and the union making a “final” demand of
a 120 per hour increase. Therefore, both sides have agreed to let an impartial arbitrator set
the wage increase somewhere between 80 and 120 ` per hour (inclusively).The arbitrator
has asked each side to submit to her a confidential proposal for a fair and economically
reasonable wage increase (rounded to the nearest dime). From past experience, both sides
know that this arbitrator normally accepts the proposal of the side that gives the most from
its final figure. If neither side changes its final figure, or if they both give in the same amount,
then the arbitrator normally compromises halfway between 100 ` in this case). Each side
now needs to determine what wage increase to propose for its own maximum advantage.
Formulate this problem as a two-person, zero-sum game.
6. Two manufacturers currently are competing for sales in two different but equally profitable
product lines. In both cases the sales volume for manufacturer 2 is three times as large as that
for manufacturer 1. Because of a recent technological breakthrough, both manufacturers will
be making a major improvement in both products. However, they are uncertain as to what
development and marketing strategy to follow. If both product improvements are developed
simultaneously, either manufacturer can have them ready for sale in 12 months. Another
alternative is to have a “crash program” to develop only one product first to try to get it
marketed ahead of the competition. By doing this, manufacturer 2 could have one product
ready for sale in 9 months, whereas manufacturer 1 would require 10 months (because of
previous commitments for its production facilities). For either manufacturer, the second
product could then be ready for sale in an additional 9 months.
For either product line, if both manufacturers market their improved models simultaneously,
it is estimated that manufacturer 1 would increase its share of the total future sales of this
product by 8 percent of the total (from 25 to 33 percent). Similarly, manufacturer 1 would
increase its share by 20, 30, and 40 percent of the total if it marketed the product sooner than
manufacturer 2 by 2, 6, and 8 months, respectively. On the other hand, manufacturer 1 would
lose 4, 10, 12, and 14 percent of the total if manufacturer 2 marketed it sooner by 1, 3, 7, and
10 months, respectively. Formulate this problem as a two-person, zero-sum game, and then
determine which strategy the respective manufacturers should use according to the minimax
criterion.