GAMES AND STRATEGIES 2024

Week 7: Cooperative games

• Discuss some remarkable examples of TU games: Seller and buyer, bankruptcy, MCST,
weighted majority, peer game.

• Analyze TU games and simple TU games.

• Find winning coalitions, veto players and null players in simple TU games.

• Understand the notion of (super/sub)additivity.

Exercise 1. Consider a game with set of players N = {1, . . . , 5}. Compute the number of all
possible non-empty coalitions for such a game without listing them.

Exercise 2. A seller (that we denote by player 1) has shop of vintage shirts and there are
two potential buyers for a specific shirt (to which we refer as player 2 and 3). Such a shirt
is evaluated 10 euro by the seller, while players 2 and 3 evaluate the product 15 and 20 euro,
respectively. The evaluations are private informations and hence, when the seller meets a buyer
with a higher evaluation of the shirt (with respect to his own evaluation), he will sell the shirt
at the higher price.
List all possible coalitions and write the characteristic function.

Exercise 3. In the bankruptcy game there is an external agent whose capital is E and there
are n players, referred to as creditors, who claim credit from the external agent. We denote by
P
ci the credit claimed by player i. The bankruptcy condition is then E < i∈N ci = C, that is
the external agent is not able to pay all creditors.
Note that we can identify such a game with the triple B = (N, c, E), where N = {1, . . . , n} is
the set of players (creditors) and c = {c1 , . . . , cn }.
We have two different ways to define a TU game for this situation: one pessimistic and one
optimistic. For any coalition A ⊆ N the two characteristic functions are given by
 
X
vP (A) = max 0, E − ci  , ∀A ⊆ N,
i∈N \A

for the pessimistic game and

!
X
vO (A) = min E, ci , ∀A ⊆ N,
i∈A

for the optimistic one. We can understand vP (A) as the sum that the coalition A will receive
from the external agent in the most adverse circumstances. This happens when the external
agent settles payments with players in N \ A before paying those in A. Conversely, vO (A)
represents the amount that the coalition A could receive under the most favorable conditions. In
this case, the external agent would settle payments with players in A before those in N \ A.
Consider the following setting: (N = {1, 2}, c = (3, 4), E = 5).

(a) List all possible coalitions and write the characteristic function for the pessimistic game
and for the optimistic one.

(b) Show that the game with characteristic function vP is superadditive, while the one with
vO is not.

1
 Figure 1

Exercise 4. Consider the graph in Figure 1 where O is a power plant that provides electricity to
the houses in N = {1, 2, 3}. The numbers on the edges are the costs of connecting two different
houses or one house with O.

(a) List all possible coalitions.

(b) How can we represent the cost of each coalition of houses? Write the characteristic func-
tion keeping attention to the fact that is a cost game.

(c) Check that the game is subadditive.

Exercise 5. Consider the simple TU game1 (N, v) where N = {1, 2, 3, 4} and ∀A ∈ 2N ) the
characteristic function v is defined as
(
1 if |A| ≥ 3 and {1, 2} ⊂ A ,
v(A) =
0 otherwise .

(a) Write the winning coalitions.

(b) How many non-empty coalitions are losing?

(c) Which are the veto players of the game?

(d) Which are the null players of the game?

Exercise 6. Consider the following game. There are 4 parties: 1,2,3,4. 1 has the 40% of votes,
2 the 23%, 3 the 19% and 4 the 18%. In order to take a decision, the majority of at least 51%
of votes is needed.

(a) Write the game as a weighted majority game and write the characteristic function.

(b) List all the winning coalitions.

(c) Is there any veto player?

(d) Is there any null player?

1
Recall that a simple TU game is a TU game in which the characteristic function v takes only values 0 or
1, i.e. v(S) ∈ {0, 1} for any S ⊂ N .

2
 Figure 2

Exercise 7. We define the peer game as follows. Let N = {1, . . . , n} be the set of players
and let T be a directed rooted tree with vertex set N . Each agent i has an individual potential
ai which represents the gain that player i can generate if all his ancestors cooperate with him.
For any agent i, denote by S(i) the set of all agents in the unique directed rooted path connecting
1 to i in T . Then, we define the peer game as the game with characteristic function
X
v(S) = ai ,
i∈N :
S(i)⊂S

that is, in the above summation we sum only ai for the agent i ∈ S such that all his ancestors
are in S.
Let us now consider such a game for the case described in Figure 2. Compute the characteristic
function knowing that (a1 , a2 , a3 , a4 , a5 ) = (6, 4, 3, 2, 1).

Exercise 8. Consider the following TU game with set of players N = {1, 2, 3} and with char-
acteristic function
v({1}) = 0 , v({2}) = 3 , v({3}) = 1 ,
v({1, 2}) = 4 , v({1, 3}) = 2 , v({2, 3}) = a ,
v(N ) = 6 ,
where a ∈ R. Find the values of a ∈ R for which such a game is superadditive.

Exercise 9. Let (N, v) be the TU-game defined as follows: N = {1, . . . , n} and for A ⊆ N :
(
1 if |A| is odd
v(A) =
|A| otherwise.

For which n ≥ 2 is the game is superadditive?