Você está na página 1de 18

Exercise sheets micro-economy game theory Knigstein

Exercise sheet 1
Exercise 1.1
Give one example of a game-like situation from your everyday life. Identify the players, the nature
of interaction, the strategies available, and the objectives that each player is trying to achieve.
Teamwork:
Group:
Interaction:
Strategic:
Rational:

Students jointly preparing a case study


If I put little effort in the project someone else must work harder to get the work done
Estimating the likelihood of free-riders in the group
Comparison of the benefits of extra work (better grade) against the costs of extra
work (less leisure time)
R&D expenditures
Group:
e.g., pharmaceutical companies
Interaction:
The first developer of a drug makes the most pro fits (thanks to the
associated patent)
Strategic & rational:
Choosing how much to spend involves thinking about competitors
commitments and possible reactions to own decision
Exercise 1.2
By carefully examining the components of a game situation group, interaction, strategy, and
rationality discuss whether the following situations qualify as a game.
a) Purchasing a house
b) A trial by a jury
a) Purchasing a house
group: seller, buyers different situation of buying a house Many buyers
achieve, objectives: buyer: buy a house; buyers wants to minimize price
seller: objective of the seller is to get the best price
interact in a strategic sense?
Individual
bargain? buying a house is usually speci fic, small,
thin market
Strategic game problem
do i want to influence the price? Are they other buyers?
b) A trial by a jury
-Players: the jury members
-strategy of each jury member: penalty or range to penalty (1 or 2 or 3 years in jail)
you might as well say: this is a suggested decision yes or no
specific penalty or just choice between yes or no there are different types
-goal of a jury member: come up with a justice decision neutral, unbiased decision
-nature of interaction more decision of individual choice or more an anticipate what other will
choose interaction is possible decision making in a group
both activities can be modelled
to remember: 4 categories:
a) players
b) players choice sets
c) goals
d) nature of strategic action
Exercise 1.3
Consider two persons. Person 1 has altruistic preferences. This means, she does not care only for
her own income but also for the income of person 2. Precisely, she attaches to each unit of own
income the same value as for two units of person 2s income. For example, she is indifferent
between two actions a and b when action a generates an income of 1 for her and 0 for person 2 and
action b generates an income of 0 for her and 2 for person 2.

a) How do person 1s preferences order the actions c, d, e associated with outcomes (1,4),
(2,1), and (3,0) respectively, where in each case the first number indicates her own income
and the second number is person 2s income?
b) Give a payoff function consistent with these preferences.
a) order the actions:
(3,0) > (2,1)
(1,4) ~ (3,0) > (2,1)
preference sign not equal ~ but indifferent
<, >, ~ are preference operators
b) they are many payoff functions
payoff vector: (x,y)
payoff x = payoff from p1
payoff y = payoff from p2
utility function = u(x,y) = x + y
u(3,0) = 3 + * 0 = 3
u(2,1) = 2 + * 1 = 2,5
u(1,4) = 1 + * 4 = 3
u (x,y) = (x + y)2
= [u(x,y)]2
monotonic transformation does not matter
W(x,y) = a + b u(x,y)
positive linear transformation with b > 0
this transformation preserves the order; it depends on a positive b
Exercise 1.4
A decision makers preferences over the action set A={a,b,c} are represented by the payoff function
u for which u(a)=0, u(b)=1, and u(c)=4.
1. a) Are the preferences of this DM also represented by the payoff function v for which v(a)=1, v(b)=0, and v(c)=2?
2. b) How about the function w for which w(a)=w(b)=0 and w(c)=8?
u(a) = 0
u(b) = 1
u(c) = 4
a) v(a) = -1; v(b) = 1; v(c) = 2
yes, because it preserves the order this order represents the same references
b) no, because we have the same payoffs for a and b
Exercise 1.5
Determine whether each of the
games differs from the Prisoners
Dilemma as introduced in the
lecture only in the names of the
actions or whether it differs also in
one or both of the players preferences.
LEFT
PD-Game has certain solution aspects: individual rationality, collective rationality, dominant
strategies, but what constitutes a PD GAME?
choice options?
Preference certain preference
For a game to be represented as a PD game it must be a 2 by 2 game with a specific order of

payoffs: example for PD:


p1

p2
a1
b1

a2
2,2
3,0

b2
0,3
1,1

preference of p1 is: according strategy vector is:


(b1,a2) > (a1,a2) > (b1,b2) > (a1,b2)
preference of p2 is:
(a1,b2) > (a1,a2) > (b1,b2) > (a2,b3)
if the preferences are like this, it will be a PD game
SOLUTION
=> left game is not a PD game, since preference ordering differs
die unteren beiden haben fr p2 beide den value 1 und deshalb ist der dann indifferent und
es gibt nicht die gleich Anordnung der Prferenzen, wie bei einem klassischen PD-Game
=> right is a PD game since preference ordering is as in PD-game-way for both players
=> right is not symmetric, but however it is nevertheless a PD Game
notion of individual rationality is important fpr PD game
Nash equilibrium analysis: What is the best choice given some choice by the other player
suggested solution
we have to check, wether one part can benefit from deviation
Wenn p1 annimmt, dass p2 y nimmt, kann p1 dann benefit, wenn er abweicht von y auf x?
Answer: NO!
Wenn p2 proposes, dass p1 y nimmt, kann er dann benefit, wenn er auf x ausweicht?
Answer: NO!
nash solution is y,y
no one can benefit from deviation
the suggested solution is the nash equilibrium!
Exercise sheet 2
Exercise 2.1: The inhabitants of the city Hope can only choose between two TV channels: TVAlpha and TV-Beta. Each channel has to decide whether to send a sports or a show program. The
expected distribution of the viewers is given in the following game table:
a) Determine the Nash equilibrium!
b) Is it reasonable that TV-Beta considers the
broadcasting of sports events?
a)
two players:
TV-Alpha + TV-Beta
two options:
sport or show
Payoff vectors:
each channel will maximizes the percentage of viewers
it is not symmetric: ansosnten mssten gleiche Zahlenwerte bei show, sport und bei
sport und show stehen, blo halt andersrum.
we search the vector, where no
player can benefit from deviating
suppose p2 chooses sports2, than p1 takes sports arrow
suppose p2 chooses show2, than p1 takes sports arrow
sport1 and show1 = Sp1 and Sh1
Sp1 = BR1 (best reply choice1)
(Sp2) Best reply function of p1
Sp1 = BR1 (Sh2)
horizontal arrow for p2
vertical arrow for p1
Sh2 = BR2 (Sp1)
Best replied function of p2

Sh2 = BR2 (sh1)


(Sp1, Sh2) unique nash equilibrium of the game!
b) no, because is not part of any nash equilibrium sp2 is never part of a best replied choice
Exercise 2.2: Two people enter a bus. Two adjacent cramped seats are free. Each person must
decide whether to sit or stand. Sitting alone is more comfortable than sitting next to the other
person, which is more comfortable than standing.
a) Suppose that each person cares only about his own comfort. Model the situation as a strategic
game. Is this game a Prisoners Dilemma? Find the Nash equilibrium (equilibria?).
b) Suppose that each person is altruistic, ranking the outcomes according to the other persons
comfort, but, out of politeness, prefers to stand than to sit if the other person stands. Model the
situation as a strategic game. Is this game a Prisoners Dilemma? Find the Nash equilibrium
(equilbria?).
c) Compare the peoples comfort in the equilibria of the two games!
a)
Players:
action profile set
utility functions:

P1; P2
A1: sit1, stand1
A2: sit2, stand2
Utility1, utility2
u1(.); u2(.)
u1(a1,a2); u2(a1,a2)

4 possible action profiles


(a1,a2) [(sit1, sit2); (sit1, stand2); (stand1, sit2) ; (stand1, stand2)]
Set with four elements, each element is a action vector or action profile
set of feasible action profiles!
(a1, a2) A with A = A1 x A2
three scenarios three number to characterize the preferences for a Player
Player 2
Player 1

Sit1
Stand1

Sit2
2,2
1,3

Stand2
3,1
1,1

it is a dominant solvable game,but no PD Game


efficiency: all action profiles are efficient, bis auf (stand1, stand2)
efficiency is a collective viewpoint
but just sit, sit is the equilibrium
b)

Player 2
Player 1

Sit1
Stand1

Sit2
2,2
3,0

Stand2
0,3
1,1

PD Game utility numbers


The social outcome is here (sit, sit) but the introduction of altruism isn't good
W (x1,x2)

= 1x1 + 2x2 social welfare function


Evaluate the ending of the game

= 2 * u 1 + u2
attach a weights! Attach depends on taste
sit1, sit2 => Welfare
=> 2 * 2 + 2 = 6
sit 1, stand 2
=> 2 * 3 + 1 = 7
stand 1, sit2
=> 2 * 1 + 3 = 5
stand 1 , stand2
=> 2 * 1 + 1 = 3
WHAT SHOULD HAPPENED IN THE SOCIETY?
c) comfort is actually higher, when they both individual or egoistic
Exercise 2.3: Each of two players has two possible actions, quiet and fink; each action pair results
in the players' receiving amounts of money equal to the numbers corresponding to that action pair in
the game table below.
The players are not selfish; rather, the preferences
of each player i are represented by the payoff
function mi(a) + mj(a), where mi(a) is the amount
of money received by player i when the action profile is a, j is the other player, and is a given
nonnegative number. Player 1's payoff to the action pair (quiet, quiet), for example, is 2 + 2.
a) Formulate a strategic game that models this situation in the case = 1. Is this game the
Prisoner's Dilemma?
b) Find the range of values of for which the resulting game is the Prisoner's Dilemma. For
values of for which the game is not the Prisoner's Dilemma, find its Nash equilibria.
a) This game is an Example of generalization: On top of the monetary values we can form into
utility functions
what is the utility for example for quiet1 and quiet2 ?
utility1(i)
utility2(j)
(quiet1, quiet2)
2 + 1* 2 = 4
2+1*2=4
(quiet1, fink2)
0+1*3=3
3+1*0=3
(fink1, quiet2)
3+1*0=3
0+1*3=3
(fink1, fink2)
1+1*1=2
1+1*1=2
Player 2
Player 1

quiet1
fink1

quiet2
4,4
3,3

fink2
3,3
2,2

unique nash equilibrium


dominant strategy
but no PD, because its end up with a efficient solution
b)
Player 1
q1,q2
q1,f2
f1,q2

quiet1
fink1

Player 2
quiet2
2+2 ; 2+2
3 ; 3
u1
2+*2
0+ *3
3+ *0

fink2
3 ; 3
1+; 1+
u2
2+*2
3+*0
0+ *3

f1,f2

1+ *1

1+ *1

For what values of do we get a PD game?


to have q1 as individually rational choice by p1 the following must hold, if p2 chooses q2
2 + 2 3
<=> 2 1
<=>
=> for > q1 is BR (q2)
to have q1 as BR (f2), the following must hold
3 1 +
<=> 2 1
<=> >
By symmetry we get q2 is BR(q1) and q2 is BR (f1) if >
=> if > , than q1 and q2 is nash equilibrium
dominant strategy qi
unique nash equilibrium
inefficient equilibrium
but it is not inefficient!
What constitutes a PD-GAME
FAZIT: we should not say, that this is a PD-GAME
b) new start: We have to start new, to get a PD-game:
conditions: (i) A player prefers fink to quiet, when the other Player chooses quiet:
=> 3 > 2 + 2
<=> 3 > 2 (1 +)
(ii) A Player prefers fink when the other player chooses fink:
=> 1 + > 3
(iii) each player prefers quiet, quiet to fink, fink
=> 2 + 2 > 1+
<=> 2 (1 + ) > 1 +
conditions (i), (ii), (iii) must hold simultaneously
what constitutes a game depends on the structure of the game, and the payoffs from each player
two strategies in the PD-Game
Wether there is a equilibrium interests just in the solution #noCONDITION
we can formulate conditions for each player separately
because the game is symmetrical, we can just prove one player
6 conditions which boiled down to 3, because of the simultaneously
When you want to model a social interaction, there are many ways to do
this few parameters or less parameters? That is on ones choice
possibilities for solving this 3 equations:
a) infinite numbers of solutions
b) one solution
c) no solution
(iii) it is always fulfilled (redundant)
=> 2 > 1
(i) 3 > 2 (1 +)
=> >
(ii) 1 + > 3
=> >
For < the game is a PD GAME
Suppose alternatively,
(i)'
<
(ii)'
< 1/3

(iii)' < ???


You have to pick the strongest conditions, cause this contains the other ones!
If = put in Table from 2.3.b)
Player 2
quiet2
quiet1
3,3
Player 1
fink1
3, 3/2

fink2
3/2, 3
3/2, 3/2

All action profiles are nash equilibria


Only f1, f2 is Nash quilibrium in the original table
=> no PD-GAM
If >

Player 2
quiet2

Player 1

fink2

quiet1
fink1

only q1,q2 is nash equilibria


=> no PD
Exercise Sheet 3
Exercise 3.1: (Stag Hunt Game). Philosopher Jean-Jacques Rousseau describes an example of a
group of hunters who wish to catch a stag. The hunters will succeed if they all remain sufficiently
attentive, but each is tempted to leave her post and catch a hare.
a) Consider the group of hunters comprising N = 2 persons and assume that each hunter
prefers one half of the stag to a hare. Describe the situation as a strategic game and find the
Nash equilibria. Which action profiles are collectively rational (efficient)?
b) Consider the group of hunters comprising N = 5 persons and assume that each hunter
prefers 1/5 of the stag to a hare. What problem do you see in modelling the situation as a
strategic game? Despite this problem, find the Nash equilibria.
There are group interests and individual interests
2 type of hunters hunters
action profiles: ai Ai, Ai = {stagi, harei}
A = Ai x aj = A1 x A2
(stag1, stag2)
2
(stag1, hare2)
0
(hare1, stag2)
1
(hare1, hare2)
1
Payoff1

2
1
0
1
Payoff2

Player 2
Player 1

stag1
hare2

stag2
2,2
1,0

hare2
0,1
1,1

one allocation is efficiency, when at least one player gets more without hunting 2,2!
two strict nash equilibria
is the action weakly dominated? No!
Pareto collective 2,2

payoff vs. strategie


payoff vectors collective
strategy domination individual
b)
1. Problem: Many action profiles (because we have 5 players)
a profile is a 5-dimensional vector
25 different possible profiles
2. Problem: We can not draw a 5 dimensional matrix
However: we can distinguish different types of action profiles and analyse these
(stag1, stag2,...stag5)
= (stagi) Vi = 1,2,...,5
(hare1, hare2,,hare5)
= (harei) Vi = 1, 2, , 5
two symmetric action profiles (2 types of action profiles)
and: There are asymmetric action profiles where some players i choose stagi
and some playersi choose harei
consider
(i) stagi Vi = 1,2,,5
Is this a Nash-equilibrium?
a) What is playeri's payoff given the proposed equilibrium profile (What is the proposed equilibrium payoff)
payoff according to the proposed solution (or candidate solution)
=> ui (stagj Vj = 1,2,,5) = 1/5 of the stag
b) What is the playeri's payoff from deviating
=> ui(harei, stagj Vj i)
c) Conclusion: Since the hunter prefers 1/5 of stag to 1 hare, stag i Vi = 1,25) is
Nash equilibrium (Deviation is not beneficial)
consider
(ii) harei Vi = 1,2,,5
Is this a nash equilibrium?
a) ui (harej Vj = 1,2,,5)
= hare
equilibiriums payoff
= ui#
b) ui (stagi, harej Vj I)
=0
deviation payoff
=~ui
c) conclusion: since ui# > ~ui the action profile harei Vi = 1,2,5 nash equilibrium
it is beneficial for one of the i-group or one for the j-group to deviate no?
Than it is a Nash equilibrium
Asymmetric profiles
a) action profiles:(harej, stagi, Vi,j = 1,2,,5, ij)
ui (harej, stagi) = ui# = 0
define:
~ui is i's utility from deviating to harei
~ui = harei
Player i would benefit from deviating
Thus, since a single or multiple players would benefit von
deviating the proposed action profile (all asymmetric profiles) are not
nash equilibria.
Graphic example:
All othersj
stag j
harej
playeri
stagi
2
0
harei
1
1
stag: 10
hare: 1
nothing = 0
b) uj (harej, stagei) = uj# = hare
~uj =
Exercise 3.2: Consider variants of the n-hunter stag hunt game in which only m hunters, with 2 m

< n, need to pursue the stag in order to catch it (assume there is only a single stag). A captured stag
is shared only by those hunters who catch it. Under each of the following assumptions on the
hunters preferences, find the Nash equilibria of the according strategic game and find out what
would be socially optimal if the caught stag and hares would be divided by all hunters:
a) Each hunter prefers the fraction 1/n of the stag to a hare.
b) Each hunter prefers the fraction 1/k of the stag to a hare, but prefers a hare to any smaller
fraction of the stag, where k is an integer with m k n.
3.2.a)
(i) suppose (stagi) Vi is the candidate solution
=> ui# = 1/n stag
((((( ui (stagj Vj , j = 1, , n)))))
Deviation payoff? ~ui = hare
Since ui# is > than ~ui, or ui# is prefer to ~ui , the candidate action profile Is a Nash
equilibrium
(ii) suppose (harei) Vi is the candidate solution
=> ui# = hare
Deviation payoff? ~ui = 0
(harei) Vi is Nash equilibrium
(iii) suppose k players with k < m hunt stag while the rest n-k players hunt hare:
Suppose playeri is one of the k stag hunters
ui# = 0
~ui = hare
=> playeri can benefit from deviation
=> candidate profile is not a Nash equilibrium
(iv) suppose k players hunt stag with m k < n while the rest n-k players hunt hare
a) suppose playeri is one the k stag hunters
ui# = 1/k stag
~ui = hare
since 1/n stag is preferred to hare and since 1/k stag is > 1/n stag , we conclude playeri prefers 1/k stag to hare
=> ui# > ~ui playeri (staghunter) can not benefit from deviation.
b) Check hare hunters Suppose playeri is one of the hare hunters
ui# = hare
~ui = 1/(k+1) stag
since 1/n stag is preferred to hare and since 1/k+1 stag 1/n stag we
conclude that playeri benefits from deviation
=> all asymmetric profiles with stag hunters and m k < n are not a nash eq'a
Overall, we conclude for 3.2.a) that only the 2 symmetric action profiles are Nash eq'a of the game
social optimal: m stag hunters and n-m hare hunters!
No equilibrium is the social optimal solution!
3.2.b)
symmetric:
(i) stagi Vi
=> ui# = 1/k stag
=> ~ui = hare
since n > k, (1/n) stag < (1/k) stag => playeri prefers have (~ui > ui#)
=> stagi Vi is not Nash eq'a
(ii) harei for all i => ui# = hare // ~ui = 0
asymmetric

(iii) suppose L players going for stag hunting L < m no stag hunting they get = 0 this is no
nash equilibrium Let L hunters hunt stag and L smaller m. The candidate profile is not a nash
equilibrium since a stag hunter benefits from deviating.
(iv) suppose L players going for stag hunting L = m
(stagi Vi = 1...m, harej Vj = m+1...n)
playeri = ui# = (1/m)stag
~ui = hare
(1/m) stag > 1/k (stag) > hare
1/m of a stag is preferred to 1/k of stag which is preferred to hare
=> i does not benefit from deviation
#
Player j: uj = hare
~uj = (1/m+1) stag
We know that player i prefers 1/k to hare
=> (1/(m+1))stag is strictly preferred to 1/k(stag)
iv a) (1/(m+1))stag > 1/k stag
if m+1 < k
no nash equilibrium
iv b) (1/(m+1))stag = (1/k)stag of m+1 = k
nash equilibrium
(v) suppose k stag hunters
(stagi Vi = 1 k, and harej Vj = k+an)
playeri : ui# = (1/k)stag, ~ui = hare
ui# > ~ui => playeri does not benefit from deviation
#
playerj : uj = hare, ~uj = (1/(k+1)) stag
We know (1/k)stag is strictly preferred to hare, but any smaller fraction of stag is
worse than hare. Therefore (1/(k+1)stag is worse than hare, which means player j does
not benefit from deviation
=> this profile is a nash equilibrium
(vi) suppose L > k hunters an l < n
suppose playeri is a stag hunter playeri ui# = 1/L stag; ~ui = hare
=> We know that 1/L stag > hare , but any smaller fraction of stay is worse than hare.
Thus 1/k of stag is worse than hare
=> playeri benefits from deviation
=> candidate profile is not a nash equilibrium
Collecting results of our analysis
3 types of equilibria:
- harei Vi i = n
- stagi Vi i = 1 k-1
harei Vj j = k n
- stagi Vi i = 1 k
harej Vj j = k+1 n
Exercise 3.3: (Hawk-Dove Game). Two animals are fighting over some prey. Each can be passive
(dove) or aggressive (hawk). Each prefers to be aggressive if its opponent is passive, and
passive if its opponent is aggressive. Furthermore, each animal prefers the outcome in which its
opponent is passive to that in which its opponent is aggressive. Formulate the situation as a strategic
game and find its Nash equilibria.
Players:
p1, p2
action profiles:
(agg1, agg2), (agg1, pass2), (pass1, agg2), (pass1,pass2)
preferences:
(i)
u1 (agg1, pass2) > (pass1,pass2)
prefers to be aggressive, if opponent is passive
(ii)
u2 (pass1, agg2) > (agg1, agg2)

(iii)

prefers to be passive, if the opponent is aggressive


u1 (agg1, pass2) > u2 (Agg1, agg2) and u1 (pass1, pass2) > u1 (pass1, agg2)
prefers the outcome in which the opponent is passive to that in which its opponent
is aggressive
=> complete ordering of the action profiles:
u1 (agg1, pass2) > u1 (pass1,pass2) > u1 (pass1, agg2) > u2 (agg1, agg2)
3
2
1
0
agg2
pass2
agg1
0,0
3,1
pass1

1,3

2,2

=> two strict nash equilibria


Exercise 3.4: (Public Good Game). Each of n people chooses whether to contribute a fixed amount
toward the provision of a public good. The good is provided if and only if at least k people
contribute, where 2 k n; if it is not provided, contributions are not refunded. Each person ranks
outcomes from best to worst as follows:
(i) any outcome in which the good is provided and she does not contribute,
(ii) any outcome in which the good is provided and she contributes
(iii) any outcome in which the good is not provided and she does not contribute,
(iv) any outcome in which the good is not provided and she contributes.
Formulate this situation as a strategic geme and find its Nash equilibria. Is there an equilibrium in
which more than k people contribute? One in which k peaple contribute? One in which fewer than k
people contribute?
Players: n people/ persons
actions/ action profiles: ai = {contribute, not contribute} Vi = 1n
Action profile: a = (a1, a2, ,an)
Preferences for each possible action profile as described from (i) to (iv)
(1) Suppose number of 0 < m < k with m being the number of contributors
consider playeri contributing. Is this a last-reply?
No! Playeri would benefit from deviation
=> any strategic profile with m < k contributors is not a Nash equilibrium
(2) Suppose no contributors: m = 0
action profile {no ,no ,no ,nono}
not contributing is best-reply for playeri => nash equilibrium
(3) Suppose m = k
Playeri: contributing: is this a best-reply? Yes! Because if player i deviates (no contribution)
the public good is not provided
Playeri is not contribute. Is it the best reply? Yes, because the public good is provided
anyway
=> Any action profile with m = k contribute and n-k non contribute is a Nash
equilibrium.
(4) m > k a contributori would benefit from deviation since the public good would still be
provided and i prefers not to contribute then.
=> There exist no Nash equilibrium with m > k contributors
Exercise sheet 4
Exercise 4.1: Consider a two-player strategic game in which each players set of actions is the set

of the players payoff functions are u1(a1, a2) = a1(a2 a1) and
a) Find the best response functions.
b) Find the Nash equilibria.
a)
u1 (a1, a2) = a1(a2 a1)
Vai Ro
u2 (a1, a2) = a1(1 a1 a2)
i = 1,2
Max u2 (a1, a2) best response function player2
a2
derivative
du1(a1,a2) / da1 = 0
<=> a2 - 2a1 = 0
<=> 1 = a2
Best response generic on a2 Best response function
2
d u1(.) / da21 = -2 < 0
1 is a maximal point (best reply)
du2(a1,a2)/ da2 = 0
<=> 1- a1 = 2a2
<=> 2 = a1
Best response function of player2
d2u2 (.) / da22 = -2 < 0
2 is a maximal point (best reply)
b) => Condition (1) and (2) must hold simultaneously
=> (1) into (2):
a2 = * (a2)
<=> a2 = a2
<=> a2 # a2 =
<=> a2 (1+ ) =
<=> a2 * 5/4 =
<=> a2# = 4/5 * = 2/5
a2# into (1) : a1# = a2# = * 2/5 = 1/5
Nash equilibrium = (a1#, a2#) = (1/5,2/5)
Equilibrium payoff:
u1# = u(a1#,a2#) = 1/5 (2/5 1/5) = 1/25
u2# = u(a1#,a2#) = 2/5 (1 1/5 2/5) = 4/25
Exercise 4.2: (Cournot Duopoly Game). Two firms compete in a market by producing and selling
homogenous goods. Let qi with i = 1, 2 represent individual quantities and let the market price p per
unit of output sold be determined according to the following function (inverse demand function):
p = 36 q with q representing the sum of both firms quantities (industry output). Let firm is
production cost per unit of output be equal to 1. Suppose each firms goal is to maximize profit i.
a) Model the given situation as a strategic game.
b) Determine and draw the best-response functions.
c) Dermine the Nash equilibria. In additon compute industry output, market price and firms
profits.
d) Suppose the two firms form a cartel: They coordinate on individual outputs and share the
sum of profits. Determine the Nash equilibria of the cartel. In additon compute industry
output, market price and firms profits.
e) Determine the firms benefits from forming a cartel according to the percentage increase
in profits. Also, compute the percentage changes in price and quantity due to the cartel.
What does this mean for consumers?

a)
Players:
action/ action profiles:
preferences/utilities:

2 firms
q1 Ro+ , q2 R0+ (actions); (q1,q2) action profile
profit
profit = Revenue minus cost
Cost = 1 * quantitiy unitcost
1 = p * q1 1 * q1
2 = p * q2 1 * q2
with p = 36 q and
q = q1 + q 2
(1)
(2)
use in (1) and (2)
1
= (36 -q1 q2) * q1 q1
= 36 q1 q2 q2q1 q1
2
= (36 -q1 q2) * q2 q2
= 36 q2 q2 q1q2 q2

b) max 1 d1/dq1 = 0
<=> 36 2q1 - q2 -1 = 0
<=> 35 -q2 = 2q1
<=>
<=> ^q1 = 35/2 q2 best response function player1
maximize ~u2 d2 (.) / dq2 = 0
<=> 46 q1 2q2 1 = 0
<=> 35 q1 = 2q2
<=> ^q2 = 35/2 q1 Best response functions player2
q2 35

^q1: ^q1 = 0 if q2 = 35
q1 = 35/2 if q2 = 0
^q2: if q1 = 0 ^q2 = 35/2
if q2 = 0 => q1 = 35

35/2

35/2

35

q2

c) solve (3) and (4) by plugin (3) in (4)


=> q2 = 35/2 (35/2 * q2)
=> q2 1/4 q2 = 35/4
=> q2 3/4 = 35/3 = 4/3
=> q2# = 35/3
=> q2# into (3)
q1 = 25/2 * 35/3 = 35/2 35/ 6
<=> q1 = 2 * 35 / 6 = 35 /3
#
total output q = q1# + q2# = 70 /3
market price: p# = 36 q# = 38/3
profits 1# = p# * q1# - q1# = 38/3 * 35/3 35/3 = 1225/9 = ungefhr 136,1
1# = 2#
d)
= 1 + 2 = p * q q
= (36) q) * q - q
= 35q q2
d/dq = 0 <=> 35 2q = 0 <=> q## = 35/2
total output that maximizes cartel profit

p## = 36 q## = 36 35/2 = 37/2


# = 37/2 * 25/2 35/2 = 1225/4 = 206,25
=> Each firm sells qi = q# = 35/4
#1 = # = 1225/8 = 153,125
e) Percentage increase in individual profit due to cartel
1##/1# = 153,1/ 136,1 = 1,125 Percentage increase = 12,5%
% increase price: p## - p# / p# * 100 = 37/2 38/3 / 38/3 = 46,1 %
% increase quantitiy is about - 25%
Exercise 4.3: (Generalized Cournot Duopoly Game). Based on 4.2 generalize the Cournot duopoly
model and solve for Nash equilibria by proceeding as follows:
a) Assume a general linear inverse demand function. Determine also relevant restrictions for
the parameters of the model.
b) Assume the same model as in 4.2 but now assume a quadratic cost function: ci(qi) = ai +
qi + 0.5 qi2 with the parameter ai representing fixed cost. At what level of fixed cost will
firm i quit staying in business?
Exercise 4.4: (Cournot Game With 3 Firms). Generalize the Cournot game described in 4.2 to the
case of 3 firms and dermine the Nash equilibria.
Exercise 4.5: (Cournot Game With N Firms). Generalize the Cournot game described in 4.2 to the
case of N firms.
a) Dermine the Nash equilibria.
b) Show that with increasing N the market price converges to the level of marginal cost (the
idealized result that should be achieved in perfectly competitive markets).
nonnegative numbers and u2(a1, a2) = a2(1 a1 a2).
Game Theory Exercise Sheet 5
Exercise 5.1: Show that the strategy pair ((2/3, 1/3), (1/3, 2/3)) is a mixed strategy Nash
equilibrium of the Bach or Stravinsky game presented
below:
We need to check whether it is rational for player 1 to
mix with (2/3,1/3), if player2 mixes with (1/3,2/3) and
vice versa.
Is (2/3, 1/3) the BR-choice given the choice of p2
(1/3,2/3)
check p1: p1 can rationally mix between actions Bach and Stravinsky only if her
expected payoff from choosing Bach is the same as the expected payoff from
choosing Stravinsky
=> EU1 (Bach1, 2) = EU1 (Str1, 2)
This must hold!
EU1 (Bach1, 2)
= 1/3 * 2 + 2/3 * 0 = 2/3
EU1 (Str1, 2)
= 1/3 * 0 + 2/3 * 1 = 2/3
=> condition (#) is fullfilled => p1 can rationally mix => Thus 1 = (2/3,1/3) is
a best-response (any mixture is a best response including the proposed mixture (2/3,1/3))
chek p2: Mixing rationally requires
=> EU2 (1, Bach2) = EU2 (1, Str2)
EU2 (1, Bach2)
= 2/3 * 1 + 1/3 * 0 = 2/3
EU2 (1, Str2)
= 2/3 * 0 + 1/3 * 2 = 2/3
=> Together the arguments established the propsed equilibrium

=> two pure equilibria (arrows) and one equilibria in mixed strategies
Exercise 5.2: Players 1 and 2 each choose a positive integer up to 4. If the players choose the same
number, then P1 pays 1 to P2; otherwise no payment is made. Each players preferences are
represented by the expected monetary payoff.
a) Represent the game in game matrix.
b) Show that the game has a mixed strategy Nash equilibrium in which each player chooses
numbers 1 to 4 with probability 1/4.
c) Assume a generalization such that every player may choose a positive integer up to K.
Show that the game has a mixed strategy Nash equilibrium in which each player chooses
each positive integer up to K with probability 1/K.
a)

player2
b)

1
2
3
4

player1
2
0,0
-1,1
0,0
0,0

1
-1,1
0,0
0,0
0,0

3
0,0
0,0
-1,1
0,0

4
0,0
0,0
0,0
-1,1

p2 mixed strategy: (q1 ,q2 ,q3 ,q4) = q


p1 mixed strategy: (p1 ,p2 ,p3 ,p4) = p

check best-reply condition fr p1


EU1 (1, q) = EU1 (2, q) = EU1 (3, q) = EU1 (4, q)
EU1 (1, q)
= -1 * + 0 * + 0 * + 0 * = -
EU1 (2, q)
= 0*1*+0*+0*
=-
EU1 (3, q)
=
=-
EU1 (4, q)
=
=-
=> We have established that p1 can mix rationally between actions 1 to 4 and p = (,
, , ) given q (, , , )
It is the BR for p1!!!
check best-reply condition fr p2
EU2 (p, 1) = EU2 (p, 2) = EU2 (p, 3) = EU2 (p, 4)
EU2 (p, 1)
=*1+*0+*0+*0
=
EU2 (p, 2)
=*0+*1+*0+*0
=
EU2 (p, 3)
=
=
EU2 (p, 4)
=
=
=> This establishes the proposed nash-equilibrium (in mixed strategies)
c)

player1

player2

1
2

1
-1,1
0,0

0,0

2
0,0
-1,1

0,0

4
0,0
0,0

-1,1

player1 chooses (p1 ,p2 ,,pK)


= p with pi = 1/K Vi = 1K
Player2 chooses (q1 ,q2 , ,qK) = q with qj = 1/K Vj = 1K
Check best-reply condition fr player1:
EU1 (a1, q) = -1 * 1/K + 0 * 1/K = -1/K
Va1 A1 (Va1 = 1,2,,K)
=> Player1 can rationally mix (1) and thus p1 0 1/K V I = 1,2, , K is a best

reply (2)
Check best-reply condition for player2:
EU2 (p, a2) = 1 * 1/K = 1/K
Va2 A2 (Va1 = 1,2,,K)
=> Player2 can mix rationally (1) and thus qj = 1/K V i= 1,2,,K is a best
reply (2)
Analysis gives two answers, two conclusions (1)+(2)
=> Thus the proposed strategy profile is a Nash equilibrium in mixed strategies.
Additional exercise:

PLAYER1

PLAYER2
T
M
B

L
-, 2
-,-,4

C
3,3
0,5,1

R
1,1
2,0,7

Is = ((,0,), (0, 1/3, 2/3) a Nash-equilibrium? (P1 M has probability = 0 )


Check Player1:
EU1 (T, q)
= 0 * ? + 3 * 1/3 + 1 * 2/3 = 5/3
EU1 (M, q) = 0 * 1/3 + 2 * 2/3
= 4/3
EU1 (B, q)
= 5 * 1/3 + 0 * 2/3
= 5/3
(T and B must be equal and the expected utility has to be smaller or equal
to T and B. It does not destroy our candidate solution, because player one can
rationally mix between B and T. And this mixture is a Best-reply)
=> Player1 can mix rationally (1) and the proposed mixture is (also) a bestreply ((,because any other mixture between T and B will also be a Bestreply))
Check Player2:
What exactly do we have to check?
(i) EU2 (p,C) = EU2 (p,R)
(ii) EU2 (p,L) EU2 (p,C)
EU2 (p, L)
= 2 * + 4 * 1/4
= 10/4
(ii) holds
EU2 (p, C)
=3*+1*
= 10/4
EU2 (p, R)
=1*+7*
= 10/4
(i) holds
=> proposed solution is a Nash-equilibrium

((( zero sum games Chess one part wins, one part looeses
constant sum games election
non constant sum game coordination games: multiple nash-equilibria. Some of them are
preferred by some players. Prediction what rational players will do Nash-equilibrium but if
you have multiple nash-equilibrium, the best-reply will not help you to conclude which action you
have to use PARETO Some nash-equilibria are social better than other nash-equilibrium.)))

Exercise 5.3: (A coordination game.) Two people can perform a task if, and only if, the both exert
effort. If they do so, they are better off than if nobody exerts effort. The worst outcome for a person
is reached if she exerts effort while the other person doesnt. Let the game be represented as below
with c being a parameter for cost (disutility) of effort with 0 < c < 1:
a) Determine both player best-response correspondences, show them in a graph and determine
all mixed strategy Nash equilibria.
b) How do the equilibrium choices vary with c?
two pure strat. equilibria
Def.:
q
1-q
p
1-p

= prob (no effort2)


= prob (effort2)
= prob (no effort1)
= prob (effort1)
player1 mixes rationally if
EU1 (no effort1, Q) = EU1 (effort1, Q)
<=> 0 * q + (-c) * (1-q) = -c * q + (1-c) * (1-q)
<=> 0 = 1 q c
=> Player1 mixes rationally if q = to 1 c

12.1.15
if q < 1-c
<=> 0 < 1 q c
<=> EU1 (no effort1,Q) < EU1 (effort1,Q)
=> p1 always chooses effort1 , i.e. , chooses effort1, with probability = 1 or 1 p = 1
or p = 0 best-reply-choice p = 0
If q > 1 c
<=> 0 > 1 q c
<=> EU1(no effort1,Q) > EU1 (effort,Q)
=> Player1 chooses no effort with p = 1 best reply choice i p = 1
best reply correspondence for player1 1is
0
if q < 1 c
p=
p [0,1]
if q = 1 c
1
if q > 1 c
simi... fr player 2 (symmetry):
best-reply-correspondence player 2:
(1,0),
if p > 1 c
B2 (p, 1-p) =
(q, 1-q)
if p = 1 c
(0,1)
if p > 1 c
B1(q,1-q) =

(1,0)
(p,p-1)
(0,1)

if q < 1 c
if q = 1 c
if q > 1 c

3 equilibria (2 are indifferent pure nash equilibria (1 is mixed)


b) formally nash equilibria (1,0) , (1,0)
probability no effort
both choose no effort
(0,1) (0,1)

(1-c,c) (1-c,c)
(B1) (B2)
Pure Equilibrium dont ??? with c. just mixed strategy
Correction EX 3.2.b) subcu)e : m+1 = k

Você também pode gostar