OutlineIn-Class Experiment on a Coordination Game

Test of Equilibrium Selection I :Van Huyck, Battalio, and Beil (1990)

Test of Equilibrium Selection II :Van Huyck, Battalio, and Beil (1991)

Test of Equilibrium Selection III : Copper, DeJong, Forsythe, and Ross (1990)

From Unique Equilibrium Multiple Equilibria

pBC, Centipede Game Unique Nash equilibriumPeople do not play the unique Nash equilibrium

Every strategy is a Nash equilibrium (i.e., Nash does not produce a sharp prediction)

The Weakest-Link Game

cbacebeea),e(e in-ii 0;0 ;)],....,[min( 1

}e,...,,{e(e,...,e)mequilibriuNash a is

21 with tuple-nAny

n players

Strategy space =

e,...,, }21{

Game A: a =$0.2, b=0.1, c=$0.6

cbacebeea),e(e in-ii 0;0 ;)],....,[min( 1

Hypotheses: Deductive vs. Inductive Principles

Payoff Dominance

Security (Maximin}

History dependentFor t > 1, minimum (t) = minimum (1) =

ee n ))1(),....,1(min( 1

Game B: a=$0.2, b=$0.0, c=$0.6

cbacebeea),e(e in-ii 0;0 ;)],....,[min( 1

Experimental Design

* Only minimum was announced after every round

Hypotheses

Payoff Dominance: {7, …, 7} in A and B

Security (Maximin}: {1,…, 1} in A but not in B

For t > 1, minimum (t) =

ee n ))1(),....,1(min( 1

Results of Treatment A

Results of Treatment A

Results of Treatment B and A’

Results of Treatment C: Fixed Pairings

Results of Treatment C: Fixed Pairing

Experimental Design

* Only minimum was announced after every round

Results of Treatment C: Random Pairings

Full Distribution of Choices

Summary

• The presence of strategic uncertainty (2 possible equilibrium selection principle) results in coordination failure and inefficient outcome

• The first-best outcome of payoff-dominance is unlikely, both initially and with repeated plays

• With repeated plays, subjects converge on secure but the most inefficient equilibrium