You are on page 1of 6

# 14.

Lectures 15-18
Muhamet Yildiz

## Cournot with Incomplete Information

Demand:

P (Q) = a Q

where Q = q1 + q2 .
The marginal cost of Firm 1 = c1 ; common knowledge.
Firm 2s marginal cost:

cH

with probaility ,

cL

with probaility 1 ,

## its private information.

Each firm maximizes its expected profit.
How to find the Bayesian Nash Equilibrium?
Firm 2 has two possible types; and dierent actions will be chosen for the two dierent
types.
{q2 (cL ), q2 (cH )}

Suppose firm 2 is type high. Then, given the quantity q1 chosen by player, its problem is
max(P cH )q2 = [a q1 q2 cH ] q2 .
q2

Hence,
q2 (cH ) =

a q1 cH
2

## Similarly, suppose firm 2 is low type:

1

(*)

max [a q1 q2 cH ] q2 ,
q2

hence
a q1 cH
.
2
Important Remark: The same level of q1 in both cases. Why??
q2 (cL ) =

(**)

Firm 1s problem:
max [a q1 q2 (cH ) c] q1 + (1 ) [a q1 q2 (cL ) c] q1
q1

[a q2 (cH ) c] + (1 ) [a q2 (cL ) c]
2

## Solve *, **, and *** for q1 , q2 (cL ), q2 (cH ).

q1 =

q2 (cH ) =

(***)

a 2cH + c (1 )(cH cL )
+
3
6

q2 (cL ) =
q1 =

a 2cL + c (cH cL )
+
3
6

a 2c + cH + (1 )cL
3

## Auctions Two bidders for a unique good.

vi : valuation of bidder i.
Let us assume that vi s are drawn independently from a uniform distribution over [0, 1]. vi
is player is private information. The game takes the form of both bidders submitting a bid,
then the highest bidder wins and pays her bid.
Let bi be player is bid.
vi (b1 , b2 , v1 , v2 ) = vi bi if bi > bj
vi bi
2

if bi = bj

0 if bi < bj
1
max(vi bi )P rob{bi > bj (vj )|given beliefs of player i) + (vi bi )P rob{bi = bj (vj )|...)
bi
2
Let us first conjecture the form of the equilibrium: Conjecture:
equilibrium
2

## Symmetric and linear

b = a + cv.
Then, 12 (vi bi )P rob{bi = bj (vj )|...) = 0. Hence,

bi

(vi bi )P rob{vj

bi a
(bi a)
} = (vi bi )
c
c

FOC:

vi + a
if vi a
2
=a
if vi < a

bi =

(1)

## The best response bi can be a linear strategy only if a = 0. Thus,

1
bi = vi .
2
Double Auction Simultaneously, Seller names Ps and Buyer names Pb . If Pb < Ps , then no

Pb +Ps
2 .

## Valuations are private information:

Vb uniform over (0, 1)
Vs uniform over (0, 1) and independent from Vb
Strategies Pb (Vb ) and Ps (Vs ).

Pb + Ps (Vs )
: Pb Ps (Vs ) =
maxE Vb
Pb
2

## Pb + E[Ps (Vs )|Pb Ps (Vs )]

max Vb
P rob{Pb Ps (Vs )}
Pb
2

where E[Ps (Vs )|Pb Ps (Vs )] is the expected seller bid conditional on Pb being greater than

Ps (Vs ).

## Similarly, the sellers problem is

Ps + Pb (Vb )
Vs : Pb (Vb ) Ps ] =
Ps
2

## Ps + E[Pb (Vb )|Pb (Vb ) Ps

max
Vs ] P rob{Pb (Vb ) Ps }
Ps
2
maxE

Equilibrium is where Ps (Vj ) is a best response to Pb (Vb ) while Pb (Vb ) is a best response to
Ps (Vs ).
There are many Bayesian Nash Equilibria. Here is one.
Ps = X

if

Pb = X

if

Vs X

Vb X.

## An equilibrium with fixed price.

Why is this an equilibrium? Because given Ps = X if Vs X, the buyer does not want to

## trade with Vb < X and with Vb > X, Pb = X is optimal.

Vb

Vb /Vs
Efficient

0
Inefficient
lack of equilibrium

## Now construct an equilibrium with linear strategies:

pb = ab + cb vb
ps = as + cs vs ,

VS

## where ab , as , cb , and cs are to be determined. Note that pb ps (vs ) = as + cs vs i

Likewise, ps pb (vb ) = ab + cb vb i

vs

pb as
.
cs

vb

ps ab
.
cb

## Then, the buyers problem is1

pb + ps (vs )
: pb ps (vs )
maxE vb
pb
2

Z pb as
cs
pb + ps (vs )
= max
vb
dvs
pb
2
0

Z pb as
cs
pb + as + cs vs
dvs
= max
vb
pb
2
0

Z pb as
cs
cs
pb as
pb + as

vb
= max
vs dvs
pb
cs
2
2 0

cs pb as 2
pb as
pb + as

vb
= max
pb
cs
2
4
cs

pb as
pb + as pb as

vb
= max
pb
cs
2
4

pb as
3pb + as
= max
.
vb
pb
cs
4
F.O.C.:
1
cs

3(pb as )
3pb + as

vb
=0
4
4cs

i.e.,
2
1
pb = vb + as .
3
3
Similarly, the sellers problem is
1

There is somewhat simpler way in to get the same outcome; see Gibbons.

(2)

Z 1
ps + ab + cb vb
ps + pb (vb )
vs : pb (vb ) ps = max p a
vs dvb
maxE
s
ps
ps
b
2
2
cb

Z
ps + ab
ps ab
cb 1
vs +
= max 1
vb dvb
ps
cb
2
2 ps ab
cb

ps ab
ps + ab
cb
ps ab 2
vs +
= max 1
1
ps
cb
2
4
cb

ps + ab
cb ps ab
ps ab
= max 1
vs + +
ps
cb
2
4
4

ps ab
3ps + ab
cb
= max 1
vs +
ps
cb
4
4

F.O.C.

1 3ps + ab
3
ps ab
1
vs +
+
1
=0
cb
4
4
4
cb

i.e.,

3
cb
3ps + ab
vs +
+ (cb (ps ab )) = 0,
4
4
4

i.e.,
ab
3ps
cb 3
ab + cb
= + vs + (cb + ab ) = vs +
2
4
4
4
2
i.e.,
2
ab + cb
.
ps = vs +
3
3
By (2), ab = as /3, and by (3), as =

ab
3

(3)

pb =
ps =

2
1
vb +
3
12
2
1
vs + .
3
4

(4)
(5)