Lecture 5-6
Applications of Nash equilibrium
Rationalizablity & Backwards
Induction
14.12 Game Theory
Muhamet Yildiz
Road Map
1. Cournot (quantity) Competition
1. Nash Equilibrium in Cournot duopoly
2. Nash Equilibrium in Cournot oligopoly
3. Rationalizability in Cournot duopoly
2.
3.
4.
5.
6.
Bertrand (price) Competition
Commons Problem
Quiz
Mixed-strategy Nash equilibrium
Backwards induction
Cournot Oligopoly
N = {1,2,,n} firms;
Simultaneously, each firm i
produces qi units of a good at
marginal cost c,
and sells the good at price
P = max{0,1-Q}
where Q = q1++qn.
Game = (S1,,Sn; 1,,n)
where Si = [0,),
P
1
i(q1,,qn) = qi[1-(q1++qn)-c] if q1++qn < 1,
-qic
otherwise.
Cournot Duopoly -- profit
qj=0.2
Profit
c=0.2
0
qi(1- qi -qj-c)
-cqi
-0.2
(1-qj-c)/2
1-qj-c
C-D best responses
qiB(qj)
q2
= max{(1-qj-c)/2,0};
Nash Equilibrium q*:
q1* = (1-q2*-c)/2;
q2* = (1-q1*-c)/2;
q1* = q2* = (1-c)/3
q1=q1B(q2)
q*
1 c
2
q2=q2B(q1)
q1
1-c
Cournot Oligopoly --Equilibrium
q>1-c is strictly dominated, so q 1-c.
i(q1,,qn) = qi[1-(q1++qn)-c] for each i.
FOC: ( q ,K, q )
[ qi (1 q1 L qn c )]
i
n
1
qi
q=q
qi
q = q*
= (1 q1* L qn* c ) qi* = 0.
That is,
2q1* + q2* + L + qn* = 1 c
q1* + 2q2* + L + qn* = 1 c
M
q1* + q2* + L + nqn* = 1 c
Therefore, q1*==qn*=(1-c)/(n+1).
Cournot oligopoly comparative statics
P
n=1
n=2
n=3
n=4
c
1
Rationalizability in Cournot Duopoly
q2
1-c
Assume that
players are
rational.
1 c
2
q1
1 c
2
1-c
Players are rational:
q2
1-c
Assume that
players know
this.
1 c
2
q1
1 c
2
1-c
Players are rational and know
that players are rational
q2
1-c
Assume that
players know
this.
1 c
2
q1
1 c
2
1-c
Players are rational; players know that players
are rational; players know that players know
that players
are rational
q
2
1-c
Assume that
players know
this.
1 c
2
q1
1 c
2
1-c
Rationalizability in Cournot duopoly
If i knows that qj q, then qi (1-c-q)/2.
If i knows that qj q, then qi (1-c-q)/2.
We know that qj q0 = 0.
Then, qi q1 = (1-c-q0)/2 = (1-c)/2 for each i;
Then, qi q2 = (1-c-q1)/2 = (1-c)(1-1/2)/2 for each i;
Then, qn qi qn+1 or qn+1 qi qn where
qn+1 = (1-c-qn)/2 = (1-c)(1-1/2+1/4-+(-1/2)n)/2.
As n, qn (1-c)/3.
Bertrand (price) competition
N = {1,2} firms.
Simultaneously, each firm i sets a price pi;
If pi < pj, firm i sells Q = max{1 pi,0}
unit at price pi; the other firm gets 0.
If p1 = p2, each firm sells Q/2 units at price
p1, where Q = max{1 p1,0}.
The marginal cost is 0.
if p1 < p2
p1 (1 p1 )
1 ( p1 , p2 ) = p1 (1 p1 ) / 2 if p1 = p2
otherwise.
0
Bertrand duopoly -- Equilibrium
Theorem: The only Nash equilibrium in the Bertrand
game is p* = (0,0).
Proof:
1. p*=(0,0) is an equilibrium.
2. If p = (p1,p2) is an equilibrium, then p = p*.
1. If p = (p1,p2) is an equilibrium, then p1 = p2...
2. Given any equilibrium p = (p1,p2) with p1 = p2, p = p*.
Commons Problem
N = {1,2,,n} players, each with unlimited
money;
Simultaneously, each player i contributes xi
0 to produce y = x1+xn unit of some
public good, yielding payoff
Ui(xi,y) = y1/2 xi.
Stag Hunt
(2,2)
(4,0)
(0,4)
(5,5)
Equilibrium in Mixed Strategies
What is a strategy?
A complete contingent-plan of a player.
What the others think the player might do under
various contingency.
What do we mean by a mixed strategy?
The player is randomly choosing his pure
strategies.
The other players are not certain about what he
will do.
Stag Hunt
(2,2)
(4,0)
(0,4)
(5,5)
Mixed-strategy equilibrium in Stag-Hunt game
Assume: Player 2 thinks that,
with probability p, Player 1
targets for Rabbit. What is the
best probability q she wants to
play Rabbit?
His payoff from targeting
Rabbit:
U2(R;p) = 2p + 4(1-p)
= 4-2p.
From Stag:
U2(S;p) = 5(1-p)
She is indifferent iff
4-2p = 5(1-p) iff p = 1/3.
5
4.5
4
3.5
3
2.5
4 - 2p
2
1.5
1
5(1-p)
0.5
0
0.2
0.4
0.6
0.8
if p < 1/3
0
q ( p ) = q [0,1] if p = 1/3
1
if p > 1/3
BR
Best responses in Stag-Hunt game
q
1/3
p
1/3
10
Bertrand Competition with costly search
N = {F1,F2,B}; F1, F2
are firms; B is buyer
B needs 1 unit of good,
worth 6;
Firms sell the good;
Marginal cost = 0.
Possible prices P =
{1,5}.
Buyer can check the
prices with a small cost
c > 0.
Game:
1. Each firm i chooses price
p i;
2. B decides whether to
check the prices;
3. (Given) If he checks the
prices, and p1p2, he buys
the cheaper one;
otherwise, he buys from
any of the firm with
probability .
Bertrand Competition with costly
search
F2
F1
F2
High
Low
F1
High
High
Low
Low
Check
High
Low
Dont Check
11
Mixed-strategy equilibrium
Symmetric equilibrium: Each firm charges
High with probability q;
Buyer Checks with probability r.
U(check;q) = q21 + (1-q2)5 c = 5 - 4 q2 c;
U(Dont;q) = q1 + (1-q)5 = 5 - 4 q;
Indifference: 4q(1-q) = c; i.e.,
U(high;q,r) = 0.5(1-r(1-q))5;
U(low;q,r) = qr1 + 0.5(1-qr)
Indifference = r = 4/(5-4q).
Dynamic Games of Perfect
Information
&
Backward Induction
12
Definitions
Perfect-Information game is a game in which all
the information sets are singleton.
Sequential Rationality: A player is sequentially
rational iff, at each node he is to move, he
maximizes his expected utility conditional on that
he is at the node even if this node is precluded
by his own strategy.
In a finite game of perfect information, the common
knowledge of sequential rationality gives
Backward Induction outcome.
A centipede game
1
D
(4,4)
(5,2)
(1,-5)
d
(3,3)
13
Backward Induction
Take any pen-terminal node
Pick one of the payoff vectors (moves) that
gives the mover at the node the highest payoff
Assign this payoff to the node at the hand;
Eliminate all the moves and the
terminal nodes following the node
Yes
Any non-terminal
node
No
The picked moves
Battle of The Sexes with perfect information
1
2
L
(2,1)
2
R
(0,0)
(0,0)
(1,2)
14
Note
There are Nash equilibria that are different
from the Backward Induction outcome.
Backward Induction always yields a Nash
Equilibrium.
That is, Sequential rationality is stronger
than rationality.
Matching Pennies (wpi)
1
Head
Tail
head
tail
(-1,1)
(1,-1)
head
(1,-1)
tail
(-1,1)
15
Stackelberg Duopoly
Game:
P
N = {1,2} firms w MC = 0;
1. Firm 1 produces q1 units
1
2. Observing q1, Firm 2 produces
q2 units
3. Each sells the good at price
P = max{0,1-(q1+q2)}.
Q
1
i(q1, q2) = qi[1-(q1+q2)] if q1+ q2 < 1,
0
otherwise.
Stackelberg equilibrium
If q1 > 1, q2*(q1) = 0.
If q1 1, q2*(q1) = (1-q1)/2.
Given the function q2*, if q1 1
P
1
1(q1;q2*(q1)) = q1[1-(q1+ (1-q1)/2)]
= q1 (1-q1)/2;
0 otherwise.
q1* = .
q2*(q1*) = .
16
