Notes

By
Mr. Mwansa JJ
 GAME THEORY
A game is a formal representation of a situation in which a number of individuals interact in a setting of
strategic interdependence. Game theory provide an intuitive, conceptual introduction to the techniques
used to study oligopoly behavior and strategic competition.

Multiperson interaction can be characterized by strategic interdependence

 Each agent in such interactions recognizes that the payoff (profit/utility) he receives depends on their
own actions and actions of other agents.

 Her best actions may depend on the actions that other individuals have even future actions that they
may take or decide not to take as a result of her current actions.

 We use noncooperative game theory to analyse settings with strategic interdependence.

 Types of Games
It is useful to classify games on the basis of
(i) the timing of moves and
(ii) uncertainty about the payoffs of rivals.
In a static game each player moves once, and when a player moves she does so not
knowing the action of her rivals.
Such a game is sometimes called a strategic game. In a dynamic game, players
move sequentially and have some idea, perhaps imperfect, about what their rivals have
done; that is, players are at least partially aware of the actions taken by others so far.
Such games are often called extensive games.
In dynamic games we can distinguish between games of perfect information, where
all players know the entire history of the game when it is their turn to move, and games
of imperfect information in which at least some players have only a partial idea of the
history of the game when it is their turn to move.
Zero Sum games vs Ponzi games
 The Basic Elements of a Game
Any game has four elements. These four elements define the structure of the
game and they are as follows:
1. Players: The identity of those playing the game. We will most often use Player
and Player two or Firm 1 and Firm 2 to denote the players participating in the
game.
2. Rules: The rules of the game specify three things:
(a) the timing of all players’ moves;
(b) the actions available to a player at each of her moves; and
(c) the information that a player has at each move.
3. Outcomes: The outcome of a game depends on what each player does when it
is her turn to move. The set of outcomes is determined by all of the possible
combinations of actions taken by players.
4. Payoffs: The payoffs of the game represent the players’ preferences over the
outcomes of the game. These can also be said to be rewards or utilities derived
from playing this strategic game.
 Example of a Game: Matching Coins
Consider a simple game “matching coins” to highlight the first 3 components:
Players: assume two players 1 and 2
Rules: each player simultaneously put a coin down(head or tail up)
Outcomes–if the two coin match (either both are heads or tails up) player 1 pays K1 to
player 2 otherwise player 2 pays player.
To complete our description of this game, we need to say what the player’s preferences
are over the possible outcomes .
We describe a players’ preferences by a utility function that assigns a utility level for each
possible outcome.
The players’ utility function is also referred to as a payoff function and utility level as her
payoff.
We assume that these utility functions take an expected utility form -(Von Neumann-
Morgenstein cardinal utility function-(hence the need to study choice under uncertainty) so
that when we a consider situations with random outcomes, we can evaluate the random
We typically assume that each player’s payoff is equal to the amount of money she
gains or loses in the game.
In cases where players’ payoff are in direct opposition to one another, we have a
situation of pure conflict and refer to such games as zero-sum games. However, game
theory is not limited to this type of interactions.
For purposes of analyzing the information/elements of game theory above, it is often
formally presented in two ways:
1.Extensive form
2.Normal form
The Extensive form captures who moves when, what actions each player can take,
what players know when they move, what the outcome is s a function of the actions
taken by the players and the players’ payoffs from each possible outcome.
The extensive form relies on the conceptual apparatus known as a game tree which
mathematically is a directed graph using a simple variation of the matching pennies
game two players move sequentially rather than simultaneously
Specifically, player 1 puts her penny down (Heads up or Tails up) first
Example of Matching Coins
The game starts with an initial decision node/root node where player 1 makes her
move deciding whether to place the penny heads up or tails up.
Each of the two possible choices for player 1 are represented by a branch from his
initial decision node At the end of each branch is another decision node at which player
2 can choose between two actions -heads up or tails up after seeing player 1’s choice.
The initial decision is referred to as player 1’s decision node; the latter two as the
player 2’s decision nodes. After player 2’s move , we reach the end of the game
represented by the terminal node.
At each terminal node, we list the player’s payoffs arising from the sequence of moves
leading to that terminal node
Perfect information and information sets
In a sequential game, when it is a player’s turn to move, she is able to observe all her
rival’s previous moves -a game of perfect information
The concept of information set allows us to analyse instances where this is not the
case
The elements of an information set are a subset of a particular player’s decision
nodes.
Implying that when play has reached one of the decision nodes in the information set
and it is that player’s turn to move , she does not know which of these nodes she is
actually at.
This ignorance arise from the fact that the player does not observe something about
what transpired previously in the game.
 Information Set
In principle, we could associate player 1’s decision node with an information set
But because player 1 knows that nothing has happened before it is her turn to move,
this information set has only one member (i.e., she knows exactly which node she is at
when she moves)
We could also draw an information set at the root node -BUT we commonly ignore
information sets that contain a single node
Any decision node not connected by a dashed line are understood to be elements of
singleton information sets.
 The common Knowledge Principle

The theory postulates that all players know the structure of the game, know that their
rivals know it, know that they know it , know that it is known, ad infinitum .
In game theory models, we often assume that the model concerned is common
knowledge among players.
Hence whatever we may know or understand about the game is must be known or
understood by the players of the game.
By contrast, a player’s private information is any information that she has that is not
common knowledge among all players in the game.
 Strategies
A fundamental concept in game theory is the notion of a player’s strategy.
Definition: A strategy is a complete contingent plan, or decision rule specifying how a player will
act in every distinguishable of action circumstance in which she might be called upon to move.
The strategy must specify the action she would take at all nodes where it is her turn to act
according to the rules of the game -whether these nodes are on or off equilibrium path of play.
Nota bene (Not well) that the distinguishable circumstances to a player are represented by her
collection of information sets [which she may need to move ].
In this case, a player’s strategy amounts to specifying how she plans to move at each one of
her information sets, should it be reached during the paly of the game
Action Vs Strategy :
 A strategy is a plan of action for every decision node where a player may be called to act.
Question: why have a plan for contingencies that her own actions ensures that they never
arise?
Answer –what would happen at such never reached nodes play a big part in studying dynamic
games
 Strategies Example
 Thus, player 2 has four possible strategies: S2 = {(H,H), (H,T), (T,H), (T,T)} where (H,T) means
“Play H if player 1 plays H and play T if player 1 plays T”
 A profile of players‘ strategy choices in an I player game is represented by a vector
 S=(s1, ……., sI) where Si is the strategy chosen by player i
 The strategy profile can also presented as si= (si, s-i) where s-iis the (I-1) vector of strategies for
players other than i
 Normal Form

Definition: the normal form is the representation of a game in a game matrix, showing
strategies available for each player and along separate dimensions (column, row, etc.) of
the matrix and the outcomes and payoffs in the multidimensional cells.
Normal Form example
 Nash Equilibrium

In words, the Nash equilibrium (NE) is a configuration of strategies (one for each player)
such that each player’s strategy is the best for him , given those of the other players (this
can be a mixed or pure strategy)

In the NE, each player’s strategy choice is a best response to the strategies actually played
by the rivals.

In the Nash equilibrium, no player has an incentive to deviate from his strategy, thereby
making a Nash equilibrium a mutual best response.

In this course we cover three way od detecting the NE namely, iterative deletion of strictly
dominated strategies, Brute force or cell by cell inspection and backward induction.
 Iterated Deletion of strictly dominated strategies

We can eliminate not only strictly dominated strategies after the first deletion of strategies
but also strategies that are strictly dominated after this next deletion of strategies, and so on.

Each additional iteration requires that players’ knowledge of each others’ rationality be one
level deeper.

A player must now know not only that his rivals are rational but also that they know that he
is, and that he knows that they know that he is rational, etc.

On iteratively eliminating strictly dominated strategies, the order of deletion does not affect
the set of strategies that remain in the end.

See the example on the next slide.

Brute force method
 Backward Induction
• This is often used for extensive form games. As the name suggests, work
backwards by looking at firm B’s options then you look at firm A.
Backward induction contin..
 Duopoly Models
In this section we will focus on three major models of a duopoly market namely;
1. Courtnot Model
2. Stackelberg Model
3. Bertrand Model

 The models considered in this are static models of oligopoly pricing: competition is limited to a single period.

 The Cournot model assumes that firms compete over quantities. We consider the derivation of equilibrium,

comparative static results, and welfare implications when the number of firms is fixed and when there is free entry.

 The Bertrand model assumes that firms compete over prices. This gives rise to the Bertrand “paradox”: when

products are homogeneous and firms have constant and equal marginal costs, the competitive result that price

equals marginal cost arises even if there are only two firms in the industry.

 We demonstrate that this result is not robust to the introduction of capacity constraints and differentiated products
1. Cournot Model of Duopoly
• In a Cournot model, firms optimize with quantity
Simultaneous game –the players decide their outputs at the
same time treating the competitor’s output as fixed
Players optimize for 'best response’
Firms maximize profit taking into account beliefs about the
rival’s output
• How to set up the model
i. Two firms,
ii. with identical costs
iii.Selling identical products, and
• Step 2: Determine best response (profit max)
Profit: 1= Rr(q1) –TC(q1)
= pq1–147q1= [(339 –q2) –q1]q1–147q1
= 192q1–q2 q1 –q12
Profit Max, First order conditions:
• d/dq1=192 –q2 –2q1= 0
• q1= 96–½ q2
• This is known as firm 1’s best response function (or reaction function)
Step 3: Equilibrium will be where both conditions specified in reaction functions are
satisfied
Firm 1:q1= 96–½ q2= B1(q2)
Firm 2: q2= 96–½ q1= B2(q1)
q1= 96–½ (96–½ q1)
q1= 64 From firm 2 reaction function:
 Stackelberg Model

First mover advantage model is a sequential game, in which the firm 1 chooses output
first and firm 2 follows.

We use backward induction to solve for the SPE of this sequential game.

Begin with follower’s output –who choses q2 that maximizes own profits

Then turn to the first mover that recognizes that it can influence the follower’s actions:
Example on Stackelberg Model
Example on Stackelberg Model
Example on Stackelberg Model
 Bertrand Model
Industry consists of small number of firms - autos, airlines, radio/TV stations in small
city, Again depends on industry definition, also changes in time, e.g. emergence of
cable TV Key features: Structural interaction - choice of each affects profits of others.
Strategic interaction - all are aware of this interdependence, and this influences their
choices.
Common idea to all market structures: each firm’s choice each chooses the variables
under its control to maximize its profit.
Difference is - given what environment (things outside its control, or constraints on its
choices) ?
Perfect competition - given market price. Monopoly - given demand curve. Oligopoly -
given the choices of other firms.
Price (Bertrand) - Mail order firms simultaneously print catalogs and commit to those
prices. In reality some industries may be close to one or the other, but again must think
of the theories as organizing principles flexibly, not literally. Other dimensions in which
oligopolists compete: product design, advertising,
References
• Dr Mudenda’s PowerPoint notes (Extracts from 2018 notes)
• Geffrey Church-Industrial Organization
• Dominic Salvatore- Managerial economics