Static Games With Incomplete Information

3.
Static games of incomplete information

3.1 Bayesian games
A static bayesian game is a simultaneous-move game with
incomplete information.
Example 1: Assume it is common knowledge that two players, P1
and P2, play the prisoners dilemma with probability 2/3 and a
dierent game with probability 1/3:
P2
C D
P1
C
D
4, 4 0, 5
5, 0 1, 1
P2
C D
P1
C
D
10, 10 0, 5
5, 0 1, 1
It is also common knowledge that player 1 observes the move of
nature (that is, distinguishes which game they are playing)
Write the extensive form representation of the game.
Denition: The normal-form representation of an n-player static
bayesian game is
(A
1
, A
2
, . . . , A
n
; T
1
, T
2
, . . . , T
n
; p
1
, p
2
, . . . , p
n
; u
1
, u
2
, . . . , u
n
)
where
A
i
is the action space of player i .
T
i
is the type space of player i (player i s type, t
i
, is privately
known by player i ).
p
i
are the beliefs of player i : p
i
(t
i
|t
i
) describes i s uncertainty
about the n 1 other players, given i s own type t
i
.
Player i s payo function is u
i
(a
1
, a
2
, . . . , a
n
; t
1
, t
2
, . . . , t
n
).
The timing of the game is:
1. Nature draws a type vector (t
1
, t
2
, . . . , t
n
), where t
i
T
i
,
according to the prior probability distribution p(t).
2. Nature reveals t
i
to player i .
3. Players simultaneously choose actions: player i chooses a
i
A
i
.
4. Payos u
i
(a
1
, a
2
, . . . , a
n
; t
1
, t
2
, . . . , t
n
) are received.
When nature reveals t
i
to player i , he or she can compute the
belief p
i
(t
i
|t
i
) using Bayes rule:
p
i
(t
i
|t
i
) =
p(t
i
, t
i
)
p(t
i
)
=
p(t
i
, t
i
)
t
i
T
i
p(t
i
, t
i
)
.
When players types are independent, p
i
(t
i
|t
i
) does not depend
on t
i
but is still derived from the prior distribution p(t).
Describe the elements in the normal form of Example 1.
3.2 Bayesian Nash equilibrium
Denition: In the static Bayesian game
(A
1
, A
2
, . . . , A
n
; T
1
, T
2
, . . . , T
n
; p
1
, p
2
, . . . , p
n
; u
1
, u
2
, . . . , u
n
),
a strategy of player i is a function s
i
(t
i
) that species an action
from A
i
for each possible type t
i
T
i
.
In a separating strategy, each type chooses a dierent action.
In a pooling strategy, all types choose the same action.
Find the strategies of each player in Example 1.
In a Bayesian equilibrium, each players strategy must be a best
response (in expectation) to the other players strategies.
Denition: In a static Bayesian game
G = (A
1
, A
2
, . . . , A
n
; T
1
, T
2
, . . . , T
n
; p
1
, p
2
, . . . , p
n
; u
1
, u
2
, . . . , u
n
),
the strategies s
= (s
1
, s
2
, . . . , s
n
) form a (pure-strategy) Bayesian
Nash equilibrium if for each player i and for each of i s types
t
i
T
i
, s
i
(t
i
) solves
max
a
i
A
i
t
i
T
i
p
i
(t
i
|t
i
)u
i
(s
1
(t
1
), . . . , s
i 1
(t
i 1
), a
i
, s
i +1
(t
i +1
), . . . , s
n
(t
n
)).
That is, no player wants to change his or her strategy, even if the
change involves only one action by one type.
In a nite game (N, A
i
and T
i
nite sets), there always exists a
Bayesian Nash equilibrium, perhaps in mixed strategies.
Bayesian Nash equilibria of Example 1:
If s
2
= D, R
1
(D) = (D, D)
If s
1
= (D, D), R
2
(D, D) = D
[(D, D), D] is BNEQ

If s
2
= C, R
1
(C) = (D, C)
If s
1
= (D, C), R
2
(D, C) = C
[(D, C), C] is BNEQ

Also, by means of the expected-payo bimatrix:
J2
D C
J1
(D, D)
(D, C)
(C, D)
(D, D)
1, 1 5, 0
2/3, 7/3 20/3, 10/3
1/3, 11/3 13/3, 8/3
0, 5 6, 6
Example 2: Firm F1 decides whether to build a new plant or not,
and rm F2 simultaneously decides if entering and compete with
F1 or not. Depending on whether the building costs of the plant is
3 o 0, the payos are:
F2
E NoE
F1
B
NoB
0, -1 2, 0
2, 1 3, 0
F2
E NoE
F1
B
NoB
3, -1 5, 0
2, 1 3, 0
Let us assume p
1
is the probability F2 assigns to high cost of
building.
(NoB, B) is a dominant strategy for F1.
Given that strategy of F1, Enter is the best response for F2 if
p
1
> 1/2 and Not Enter is its best response if p
1
< 1/2.
Example 2: Now the low cost is 1.5.
F2
E NoE
F1
B
NoB
0, -1 2, 0
2, 1 3, 0
F2
E NoE
F1
B
NoB
1.5, -1 3.5, 0
2, 1 3, 0
The expected-payo matrix, depending on p
1
, is:
F2
E NoE
F1
(B, B)
(B, NoB)
(NoB, B)
(NoB, NoB)
1.5(1 p
1
), -1 3.5-1.5p
1
, 0
2(1 p
1
), 1-2p
1
3 p
1
, 0
1.5+0.5p
1
, 2p
1
1 3.5-0.5p
1
, 0
2, 1 3, 0
[(NoB, NoB), E] is BNEQ for all p
1
.
[(NoB, B), NoE] es BNEQ if and only if p
1

1
2
.

Static Games With Incomplete Information

Uploaded by

Copyright:

Available Formats

Static Games With Incomplete Information

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Static Games With Incomplete Information

Uploaded by

Copyright:

Available Formats

3.

Static games of incomplete information

[(D, D), D] is BNEQ

[(D, C), C] is BNEQ

You might also like