In practice, however, achieving the Pareto efficient outcome may be problematic
because subjects are not in a stable partnership, cannot communicate their intentions to
others, and can neither commit to nor enforce cooperation. One also wonders whether the
subject perceives the grim-trigger punishment as a plausible threat. Given these frictions,
subjects face a double challenge: not only must they be able to coordinate on the Pareto
efficient outcome, but also coordinate on a credible threat that can support continuous
cooperation. Our goal is to identify behavioral elements and institutional characteristics
that are associated to the emergence, sustainability, and breakdown of cooperation.
This paper reports the experimental results from four treatments of matching
economies where interaction is indefinitely repeated, based on a probabilistic
continuation rule. Treatments differ in two dimensions: the level of information about
action histories and the punishment technology. Under private monitoring, subjects
observed only their own history and under public monitoring, they observed the history
of the whole economy. In some treatments subjects could only punish by defecting, while
in the personal punishment treatment, they could pay a cost to inflict a loss on their
opponent.
Our study addresses the following research questions: can strangers who interact
indefinitely achieve substantial levels of cooperation and efficiency? Which institutions
for monitoring and enforcement promote cooperation? What classes of strategies are
adopted in economies that achieve high efficiency? We obtained the following results.
First, efficiency levels in our experimental economies are high and increasing with
experience, even under private monitoring; this result provides empirical support for the
theoretical findings in Kandori (1992) and Ellison (1994). Second, costly personal
punishment significantly promotes cooperation; however, not all monitoring institutions
promote cooperation. We report high cooperation levels in situations where subjects
know identities and histories of opponents (non-anonymous public monitoring) but not
when identities are unobservable (anonymous public monitoring). Finally, subjects
appear to have preferences for certain strategies. In particular, the average subject: (a)
avoids indiscriminate strategies; (b) shows a strong tendency to defect with opponents