E3001 Tutorial 2 Solutions

1. Consider a penalty shootout: The probability that the kicker will score if he kicks to the left and the
goalie jumps to the right is 0< p< 1. We want to see how the equilibrium probabilities in the mixed
strategy NE changes as p changes.

Kicker

Kick left Kick Right

Goalie Jump Left 1,0 0,1
Jump Right 1− p , p 1,0

(a) If the goalie jumps left with probability π G, what is the expected utility of the kicker if he kicks right?

The expected utility is π G ×1+ ( 1−π G ) ×0=π G.

(b) If the goalie jumps left with probability π G, what is the expected utility of the kicker if he kicks left?

The expected utility is π G ×0+ ( 1−π G ) × p=( 1−π G ) × p .

(c) Find the probability π G that makes the kicker indifferent between kicking left and kicking right (Your
answer will be a function of p.)

Setting the two equations above to be equal, we get π G= p /(1+ p)

(d) If the kicker kicks left with probability π K , what is the expected utility of the goalie if he jumps left?

π K . Method is similar as in (a).

(e) If the kicker kicks left with probability π K , then what is the expected utility of the goalie if he jumps
right.

( 1− p ) π K +(1−π K ). Method is similar as in (b).

(f) Find the probability π K that makes the expected utility to the goalie equal from jumping left or
jumping right.

1
Setting the two values above to be equal, we get π K =
1+ p
(g) What is the unique mixed strategy nash equilibrium in the game?

The mixed strategy NE is:

1
Kicker kicks left with probability .
1+ p
p
Goalie dives left with probability
1+ p
The best response graph looks something like below. As you can see, there is only 1 unique intersection.
Try and derive this yourself to see if you understand the concept!

πG

p
1+ p

πK
1 1
1+ p

You can also calculate the probability of each outcome occurring and hence the expected payoffs to
players in the mixed NE:
1 p
With probability × , the Kicker kicks left and the Goalie dives left.
1+ p 1+ p
1 1
With probability × , the Kicker kicks left and the Goalie dives right.
1+ p 1+ p
p p
With probability × , the Kicker kicks right and the Goalie dives left.
1+ p 1+ p
p 1
With probability × , the Kicker kicks right and the Goalie dives right.
1+ p 1+ p
2
p 1 p p
The expected payoffs of the Goalie are 2
.1+ 2
. ( 1− p )+ 2
.0+ .1
( 1+ p ) ( 1+ p ) ( 1+ p ) ( 1+ p )2
2
p 1 p p
The expected payoffs of the Kicker are 2
.0+ 2
. p+ 2
.1+ .0
( 1+ p ) (1+ p ) ( 1+ p ) ( 1+ p )2

(h) The variable p tells us how good the kicker is at kicking the ball into the left side of the goal when it is
undefended. As p increases, does the equilibrium probability that the kicker kicks to the left increase or
decrease? Explain why this happens in an intuitive manner.

It decreases. The better the kicker’s weak side gets, the less often the goalie defends the kicker’s good
side. So the kicker can kick to good side more often.
2. Two software companies sell competing products which are substitutes, resulting in sales being
influenced by each product’s price. Let p1 be the price and x 1 the quantity sold of product 1, and p2 and
1 1 1 1
x 2 for product 2. Then x 1=1000(90− p1 + p 2) and x 2=1000(90− p2 + p 1). Each company
2 4 2 4
has incurred a fixed cost, but the cost of selling to an extra user is zero. Therefore, each company will
maximize its profits by choosing the price that maximizes its total revenue.

(a) Write an expression for the total revenue of company 1, as a function of its price p1 and the other
company’s price p2.

1 2 1
Revenue 1=p 1 x 1=1000(90 p 1− p 1 + p1 p 2)
2 4

(b) Company 1’s best response function BR1 (p 2) is defined as the price for product 1 that maximizes
company 1’s revenue given that the price of product 2 is p2. Write down BR 1 ( p 2) .

Take the derivative of revenue with respect to p1 and solve for the revenue-maximizing p1 given p2.

(
1000 90− p1 +
1
p =0
4 2 )
1
p1=90 + p 2
4
1
BR1 ( p2 ) =90+ p
4 2

(c) Use a similar method to solve for company 2’s best response function B R 2 (p 1).

1
BR2 ( p1 ) =90+ p
4 1

(d) Solve for the Nash equilibrium prices.

The Nash equilibrium is the intersection of the companies’ best response functions:

1
p1=90 + p 2
4
1
p2=90+ p 1
4
Solving this gives:

p1=120 , p2=120
Let us try to solve for the “equilibrium” when companies move sequentially. This will be done via
backward induction which will be discussed in greater detail in lecture 3.

(e) Suppose that company 1 sets its price first. Company 2 knows the price p1 that company 1 has chosen
and it knows that company 1 will not change this price. If company 2 sets its price so as to maximize its
revenue given that company 1’s price is p1, then what price will company 2 choose?

This is just the same as before:

1
BR2 ( p1 ) =90+ p
4 1

(f) If company 1 is aware of how company 2 will react to its own choice of price, what price will company
1 choose?

Since company 1 takes into account company 2’s strategy, we substitute p2=BR2 ( p 1 ) into the revenue
function.

1 2 1
Revenue 1=p 1 x 1=1000(90 p 1− p 1 + p1 p 2)
2 4
1 1 2 45 1 2
Substituting p2=90+ p 1, then Revenue 1=1000(90 p1− p1 + p1 + p1 )
4 2 2 16
Take a derivative of revenue with respect to p1 and solve for the revenue-maximizing p1.

p1=900 /7

3.

(a) Game A: Victoria and Albert are roommates for 1 week. Each of them prefers a clean room to a dirty
room, but neither likes housecleaning. Their payoffs are as hence as follows. Derive all Nash equilibrium
of this static game A, where they decide on their actions simultaneously.

Clean Don’t Clean Game A

Albert Clean 5,5 2,6
Don’t Clean 6,2 3,3

The dominant strategy is not to clean, and hence (Don’t clean, Don’t clean) is the unique Nash equilibrium
of this game.

(b) Game B: Suppose now, in the second week, Victoria and Albert are planning to eat out on Saturday.
They are deciding whether to have Korean BBQ or Macdonalds. They both prefer Korean BBQ to
Macdonalds, especially if they have happened to have the meal together. The payoffs are hence as
follows. Derive all Nash equilibria of this static game B, where they decide on their actions
simultaneously.

Victoria
BBQ Macs
Albert BBQ 5,5 2,1 Game B
Macs 1,2 3,3

Note that all Nash equilibria refers to both pure and mixed NE.

There are 2 pure strategy Nash equilibria, (BBQ,BBQ), (Macs,Macs) and a mixed strategy NE (choose BBQ
with probability 1/5, choose BBQ with probability 1/5, ) .

(c) Sketch the extensive form of the entire game where Game B is played after Game A.
(You can ignore filling in the payoffs.)

Game B
Clean, Clean

Clean, No Clean Game B

Game A
No Clean, Clean
Game B
No Clean, No Clean

Game B
Fill in Game A and Game B as below.

Game A Game B
C BBQC

C NC BBQ Macs

C BBQ
NC Macs

NC Macs

4. In the game of “Chicken”, two players in high-speed cars drive toward each other. The possible
outcomes are as follows:

- If one driver swerves while the other does not, the swerving driver is seen as the “chicken”, and
the non-swerving driver is considered the hero.
 - If both drivers swerve, they are both seen as “chickens”.
- If neither driver swerves, both end up in the hospital as a collision occurs.

A payoff matrix for a chicken-type game is the following:

Player 2
Swerv Don’t Swerve
e
Player 1 Swerve 1,1 1,2
Don’t Swerve 2,1 0,0

(a) Does this game have a dominant strategy? Why?

No. Their best response depends on whether the other Swerves or Not.

(b) What are the two Nash equilibria in pure strategies?

The two outcomes where one teenager swerves and the other does not.

(c) Find the Nash equilibrium in mixed strategies for this game.

Note that pure strategies are also mixed strategies. The set of all mixed strategy NE are (S,DS) (DS,S) and
(swerve with probability ½, swerve with probability ½ ).

To calculate this, let Player 1’s probability of choosing swerve be p and Player 2’s probability of choosing
swerve be q.

Player 1 strictly prefers swerve to don’t swerve, i.e. p=1 if

1
q + ( 1−q ) >2 q+0 or 1>2 q . I.e. q < .
2
1
Player 1 strictly prefers don’t swerve to swerve, i.e. p=0 if q > .
2
1
Player 1 is indifferent between swerving and not swerving, p ∈[0 ,1] when q= .
2
Likewise:

1
Player 2 strictly prefers swerve to don’t swerve, i.e, q=1 if p+ ( 1− p )> 2 p+ 0 or 1>2 p . I.e. p¿ .
2
1
Player 2 strictly prefers don’t swerve to swerve, i.e, q=0 if p> .
2
1
Player 2 is indifferent between swerving and not swerving, i.e. q ∈[0 , 1] when p¿ .
2

Drawing out the best response functions, we get:

q
1

0.5
 0.5 1

Conceptual Questions

5. Write out the steps for finding (all) Nash equilibria of a two-player game with 2 actions.

To practice solving for NE in such games, i) write out two-player games with 2 actions where there are no
dominant strategies for both players, ii) solve for the NE using your steps above.

For the steps for finding (all) Nash equilibria, please see slides p8-13. Here is an example of using the
steps to find all NE:

Player 2
Action A Action B
Player 1 Action X 6,3 2,2
Action Y 2,2 3,6

Solving for Nash Equilibria:

Let's assume that Player 1 chooses Action A with probability p and Action B with probability 1−p, and
Player 2 chooses Action X with probability q and Action Y with probability 1−q.

Given the Player 2’s strategy,

- The expected payoff of choosing Action X for Player 1 is 6q+2(1-q)

- The expected payoff of choosing Action Y for Player 1 is 2q + 3(1-q)

Choosing Action X is a best response for Player 1 iif

6q+2(1-q) ≥ 2q + 3(1-q)
 q≥1/5

Choosing Action Y is a best response for Player 1 iif

6q+2(1-q) ≤ 2q + 3(1-q)

q≤1/5

The Player 1’s best response correspondence to any mixed strategy of Player 2’s q is:

- If q≥1/5, Action X is a best response. I.e., p = 1.

- If q≤1/5, Action Y is a best response. I.e., p = 0.
- If q=1/5, either Action X or Action Y is a best response. I.e., 0≤ p ≤1.

Likewise, the Player 2’s best response correspondence to any mixed strategy of Player 1’s p is:

- If p≥4/5, Action A is a best response. I.e., q = 1.

- If p≤4/5, Action B is a best response. I.e., q = 0.
- If p=4/5, either Action A or Action B is a best response. I.e., 0≤ q ≤1.

By identifying the intersection points of p and q, we can find 3 NE here: (Action X, Action A), (Action Y,
Action B), and (4/5, 1/5).

6. In the game below, which is just the matching pennies game in the lecture, with an extra strategy for
Player 1,

(a) Find a mixed strategy which strictly dominates strategy B of Player 1.

(b) Will strategy B ever be part of a Nash equilibrium?

(c) State the Nash equilibrium of this game.

Player 2
L R
T -1,1 1,-1
Player 1 M 1,-1 -1,1
B -0.1,0.5 -0.1,0.5

(a)

Let's assume that Player 1 chooses T with probability p and M with probability 1−p

B is dominated by the above mixed strategy if the following conditions are satisfied:

(If Player 2 Chooses L): p ×−1+ ( 1− p ) ×1>−0.1

(If Player 2 Chooses R): p ×1+ ( 1− p ) ×−1>−0.1

Which is 0.45< p< 0.55

(b) No, a strictly dominated strategy is never a best response.

(c) Given B is dominated, we remove strategy B and calculate the mixed strategy:

Player 2
L R
T -1,1 1,-1
Player 1 M 1,-1 -1,1

There is no pure NE in this game. The mixed NE is (p=½, q=½ ) as mentioned in the lecture.

Where Player 1 chooses T with probability p and M with probability 1−p, and Player 2 chooses L with
probability q and R with probability 1−q.