Scribd
Upload
78 views

Forward Induction: ECON2112

The document discusses the concept of forward induction in game theory. It uses examples of games like the Battle of the Sexes with entry and burning money to illustrate the idea. Forward induction says that players should make choices consistent with deductions about other players' rational behavior based on past actions. It may eliminate some subgame perfect equilibria that do not conform with this principle from consideration. The document provides a formal definition and discusses how to apply forward induction concepts to determine which SPE are reasonable in games.

Uploaded by

leobe89
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
78 views

Forward Induction: ECON2112

The document discusses the concept of forward induction in game theory. It uses examples of games like the Battle of the Sexes with entry and burning money to illustrate the idea. Forward induction says that players should make choices consistent with deductions about other players' rational behavior based on past actions. It may eliminate some subgame perfect equilibria that do not conform with this principle from consideration. The document provides a formal definition and discusses how to apply forward induction concepts to determine which SPE are reasonable in games.

Uploaded by

leobe89
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

Forward Induction

ECON2112

Forward Induction
Battle of the Sexes with Entry

1
Out

In
1
T

2,5

B
2

4, 1

0, 0

0, 0

1, 4

The game has 3 Subgame perfect equilibria:


(InT , L)
(OutB , R )
( 3 OutT + 1 OutB , 1 L + 3 R )
4
4
4
4

Are all of them reasonable?

Forward Induction
Battle of the Sexes with Entry

Let us analyze (OutB , R ).


1
Out

In
1
T

2,5

B
2

4, 1

0, 0

0, 0

1, 4

Player 1 does not move In because is he does, the strategy


combination (B , R ) will be played in the battle of the sexes.

The strategy combination (B , R ) is a rational description of


behavior in the battle of the sexes.

However, in the context of the current game, is (B , R ) a good


description of rational behavior in the battle of the sexes
subgame?

Forward Induction
Battle of the Sexes with Entry

Suppose that we propose the players to play according to (OutB , R ).


1
Out

In
1
T

2,5

B
2

4, 1

0, 0

0, 0

1, 4

Suppose that player 1 deviates and plays In.

Player 2 was not supposed to move under (OutB , R ).

How is player 2 going to react when he is called to play?

Will he stick to R?

Forward Induction
Informal Definition

Definition (Forward Induction (Informal Idea))


Players should make their choices in a way consistent with deductions
about other players rational behavior in the past.
Remember:

Definition (Backwards Induction (Informal Idea))


Players should make their choices in a way consistent with deductions
about other players rational behavior in the future.

Forward Induction
Battle of the Sexes with Entry

The equilibrium proposed is (OutB , R ).


1
Out

In
1
T

2,5

B
2

4, 1

0, 0

0, 0

1, 4

Suppose that player 2 is not supposed to move, but he is called to


move anyway. He must, therefore, conclude that player 1 has
deviated.

When thinking about what to do (whether or not to stick to R) he


has to bear in mind that player 1 is rational.

What will player 2 do?

Forward Induction
Battle of the Sexes with Entry

When player 2 has to move in the subgame he has effectively received


the following message from player 1:
Look, I had the opportunity to get 2 for sure, and
nevertheless I decided to play in this subgame, and my
move is already made. And we both know that you cannot
talk to me because we are in the game, and my move is
made. So think now well, and make your decision.

Forward Induction
Battle of the Sexes with Entry

1
Out

In
1
T

2,5

B
2

4, 1 0, 0 0, 0 1, 4
A subgame is a part of the game that could be considered as a
separate game.

However, it should not be treated as a separate game, because it


has been preceded by a very specific way of communication: the
play leading to the subgame.

Forward Induction
Battle of the Sexes with Entry

1
Out

In
1
T

2,5

B
2

4, 1

0, 0

0, 0

1, 4

If player 2 has to move, he must conclude that player 1 gave up a


payoff equal to 2 in an effort to get 4.

By forward induction, player 2 in fact knows that player 1 has


moved InT .

Player 2 will, consequently, play T .

By backwards induction, player 1 will foresee this and will play


InT .

Forward Induction
Battle of the Sexes with Entry

1
Out

In
1
T

2,5

B
2

4, 1

0, 0

0, 0

1, 4

Set of Subgame Perfect equilibria:


{(InT , L), (OutB , R ), ( 3 OutT + 1 OutB , 1 L + 3 R )}.
4
4
4
4

Set of Subgame Perfect equilibria conforming with forward


induction:
{(InT , L)}.

Forward Induction
Burning Money

1. Player 1 decides whether or not burning 2 Dollars.

2. Player 2 observes if player 1 has burnt the money.

3. Players play:

T
B

L
4, 1
0, 0

R
0, 0
1, 4

Forward Induction
Burning Money

1
Burn

Not Burn
1
T2
B2

1
T1

B1
2

L1

R1

2
L1

R1

L2

2, 1 2, 0 2, 0 1, 4 4, 1

R2

L2

R2

0, 0

0, 0

1, 4

Forward Induction
Burning Money

1
Burn

T1
B1

L1
2, 1
2, 0

R1
2, 0
1, 4

Not Burn

T2
B2

L2
4, 1
0, 0

R2
0, 0
1, 4

The game has 9 subgame perfect equilibria.

Only one of them conforms with forward induction.

Forward Induction
Burning Money

1
Burn

T1
B1

L1
2, 1
2, 0

Not Burn

R1

2, 0
1, 4

T2
B2

L2
4, 1
0, 0

R2
0, 0
1, 4

Suppose player 1 burns the money.

By forward induction, player 2 must conclude that player 1 is


playing T1 in the subgame.

Note that Burning the money and playing B1 is strictly dominated


by not burning.

If player 1 burns the money player 2 will, therefore, play L1 .

Forward Induction
Burning Money

1
Burn

T1
B1

L1
2, 1
2, 0

Not Burn

R1

2, 0
1, 4

T2
B2

L2
4, 1
0, 0

R2
0, 0
1, 4

By backwards induction, Player 1 knows that he can guarantee


himself a payoff equal to 2 by burning the money.

Suppose now that player 2 observes that player 1 did not burn the
money.

Player 2 knows that player 1 gave up a payoff equal to 2.

Therefore, by forward induction, player 2 must conclude that


player 1 is playing T2 in the subgame.

Forward Induction
Burning Money

1
Burn

T1
B1

R1
2, 0
1, 4

T2
B2

L2
4, 1
0, 0

R2
0, 0
1, 4

By backwards induction, Player 1 knows that:

L1
2, 1
2, 0

Not Burn

If he burns the money, player 2 will play T1 in the subgame.


If he does not burn the money, player 2 will play T2 in the
subgame.

Consequently, player 1 will not burn the money at the root of the
game.

Forward Induction
Burning Money

1
Burn

T1
B1

L1
2, 1
2, 0

Not Burn

R1

2, 0
1, 4

T2
B2

L2
4, 1
0, 0

R2
0, 0
1, 4

The unique subgame perfect equilibrium that conforms with


forward induction is (Not BurnT1 T2 , L1 L2 ).

(Not BurnT1 T2 , L1 L2 ) gives player 1 a payoff equal to 4 and


player 2 a payoff equal to 1.

(Having the opportunity to hurt oneself can, sometimes,


guarantee a higher payoff.)

SPE and Forward Induction


in the Beer-Quiche Game

The Beer-Quiche Game

How to determine whether a SPE equilibrium of such


a game satisfies the Forward Induction criterium?
Heres a three step procedure:
1.
2.

3.

Consider an equilibrium
For any out-of-equilibrium message m, determine the set S(m) of Player
1s types that yield a payoff strictly less than the equilibrium payoff
whatever Player 2s best response.
Player 2 can attribute a message m only to player 1s types that do not
belong to S(m). If there exists a message m such that, for any best reply
of Player 2, there exists at least one type of Player 1 which would deviate
to m, then the equilibrium considered does not satisfy the forward
induction criterium.

How to determine whether a Nash equilibrium of such a


game satisfies the Forward Induction criterium?
Application to the Beer-Quiche game (which as two Nash equilibria in pure
strategies: (beersbeerw,notbduelq) and (quichesquichew, duelbnotq).

Step 1: Consider first (beersbeerw,notbduelq)


Step 2: m = quiche, S(m) = {surly, wimp}
If m = quiche and player 1 is surly, then Player 2 best replies by choosing notq
(this will yield him a payoff of 1>0). With such a best reply from Player 2,
Player 1 earns 2 which is less than what he would earn in the equilibrium
considered in Step 1. Therefore Player 1s type surly belongs to S(m).
If m = quiche and player 1 is wimp, then Player 2 best replies by choosing duelq
(this will yield him a payoff of 1>0). With such a best reply, Player 1 earns 1
which is less than what he would earn in the equilibrium considered in Step
1. Therefore Player 1s type wimp also belongs to S(m).
Step 3: Since S(m) contains all the possible types of Player 1, Player 1 has no
type that could yield him a higher payoff given Player 2s best replies.
Therefore, the equilibrium considered in Step 1 satisfies the Forward
Induction criterium.

How to determine whether a Nash equilibrium of such a


game satisfies the Forward Induction criterium?
Application to the Beer-Quiche game which as two Nash equilibria in pure
strategies: (beersbeerw,notbduelq) and (quichesquichew, duelbnotq).

Step 1: Consider now (quichesquichew, duelbnotq)


Step 2: m = beer, S(m) = {wimp}
If m = beer and player 1 is surly, then Player 2 best replies with notb (this will
yield him a payoff of 1>0). With such a best reply, Player 1 earns 3 which is
more than what he would earn in the equilibrium considered in Step 1.
Therefore Player 1s type surly does not belong to S(m).
If m = beer and player 1 is wimp, then Player 2 best replies with duelb (this will
yield him a payoff of 1>0). With such a best reply, Player 1 earns 0 which is
less than what he would earn in the equilibrium considered in Step 1.
Therefore Player 1s type wimp belongs to S(m).
Step 3: Since surly does not belong to S(m) and since Player 1 earns a payoff of
3 given Player 2s best reply, Player 1 will deviate from having a quiche upon
being surly: he will choose to have a beer. Therefore, the equilibrium
considered in Step 1 does not satisfy the Forward Induction criterium.

You might also like