For instructors

Teaching notes

Two colleagues lose each other at a crowded railway station with no phones. Two firms would both gain from adopting the same technical standard and have no channel through which to agree on one. In such situations any common choice is as good as any other, and the difficulty lies in finding the same one.

What the game shows

The game tests whether people can coordinate without communication, and by what means. It shows that players use features of a problem that have no bearing on payoffs, such as prominence, convention and shared background, and that these features do a great deal of work that formal game theory leaves unexplained. It also shows how much coordination depends on a common culture, which makes the results of one class different from those of another.

The theory

A pure coordination game has as many equilibria as there are possible answers, and the payoffs do not distinguish among them. Schelling (1960) argued that people resolve the problem by looking for a focal point, an answer that stands out for reasons both sides can expect the other to recognize. The labels of the strategies, which standard theory treats as irrelevant, therefore become the object of the players' reasoning.

Mehta, Starmer and Sugden (1994) gave the idea experimental content. They separate the answer that first comes to a person's mind from the answer that a person expects to come to the minds of others, and from the answer selected by a rule that both can be expected to apply. Sugden (1995) provides a formal treatment in which players choose among decision rules defined over labels.

What to expect

Schelling (1960) reported informal questionnaires in which 36 of 42 respondents chose heads over tails and two fifths named the number 1 when asked for any positive number. Asked where to meet a stranger in New York, a majority of his respondents chose Grand Central Station, and nearly all chose noon.

The comparison in Mehta, Starmer and Sugden (1994) between people who merely pick an answer and people who try to match a partner is the more instructive result. Asked for a flower, 35 percent of those who picked named a rose, against 67 percent of those who tried to coordinate. Asked for a positive number, fewer than 5 percent of those who picked chose 1, and 7 was their most common answer. Among those who tried to coordinate, 40 percent chose 1. Coordination therefore rests on reasoning about salience and is something more than shared taste.

Focal points lose much of their force when interests diverge. Crawford, Gneezy and Rottenstreich (2008) found that a salient label produced a coordination rate of 64 percent when both players earned the same amount, and of 38 percent when one outcome paid one player $5.10 in place of $5.

Tips and things to look out for

References