Discussion Papers 745
5 Experimental results
dilemma game is ostracism: players who defect are expelled. Closely related to that theoreti-
cal paper, Cinyabugama, Page and Putterman (2005) show in an experiment that the threat of
costly expulsion (through majority vote in the group) can induce cooperation in a public-
goods game. While these two papers consider models that allow players to expel uncoopera-
tive others, Hauk 2003 examines a game that, perhaps more closely related to our paper, pro-
vides the opportunity to opt out of a finitely repeated prisoners’ dilemma game. She finds that
defection in early periods is punished by opting out.
In the VOLUNTARY SYM treatment, we observe a tendency (just failing significance) to
choose team remuneration after team remuneration was realized, or inertia with respect to the
choice of team remuneration.11 However, after private remuneration, there is a weak tendency
(again, just failing significance) to choose team rather than private remuneration.12 In other
words, there is some tendency not to stay with private remuneration. This suggests that the
choice of private remuneration can be a temporary threat rather than intended to be a long-run
situation.
Similarly, in the VOLUNTARY ASYM treatment, we observe a tendency to choose team remu-
neration after team remuneration was realized in the previous period. This is true for low-cost
and for high-cost players (binomial test, 10-percent significance). In the 15 cases in which
private remuneration was chosen (in ten of these cases it was the low-cost player, and in five
of these cases it was the high-cost player), the player had contributed more effort than the
other player in the previous period. After private remuneration was realized, there was no
clear-cut tendency to choose either remuneration.
Further support for our punishment hypothesis is given in the VOLUNTARY SYM treatment
by the fact that in 21 out of the 30 cases in which a participant chose private remuneration
after having experienced team remuneration in the previous period, the participant had contri-
buted more effort than the other player. In six of the groups the majority of choices of private
remuneration after team remuneration were made after the player had contributed a higher
effort than the other player. In one group the players who chose private remuneration after
team remuneration had contributed a lower effort than the other player. In another group the
11 After team remuneration, five groups chose team remuneration while only one group chose private remunera-
tion in more than half of all cases and one group chose team and private remuneration equally often. The latter
two groups account for only 6.73 percent of all the cases in which team remuneration was observed in t-1.
12 After private remuneration, five groups chose team remuneration while only two groups chose private remu-
neration in more than half of all cases and one group chose private and team remuneration equally often.
21