good games, where average cooperation in experiments is generally positive but
declining over time (Palfrey and Rosenthal, 1994). Even if one expects some degree of
cooperation in our private monitoring economies given the salience of the payoffs, the
increase in cooperation from one cycle to the next displayed in Figure 4 is in sharp
contrast with the evidence from finitely repeated games.
Result 3. In the anonymous treatments, the introduction of public monitoring did not
improve cooperation over private monitoring.
Subjects in public monitoring possess information about the choices of others that is
unavailable in private monitoring. Figure 2 shows that when this information is
anonymous, it does not foster cooperation. Average cooperation is around 59% in both
treatments, and the difference is statistically insignificant (Mann-Whitney test, p-value
0.418, n1=n2=50). First period averages lead to the same conclusion (Table 4).
Result 4. The introduction of personal punishment in the anonymous treatments
increased cooperation.
Figures 2 and 3 and Table 4 provide support for Result 4. When we add personal
punishment to economies with private monitoring, average cooperation jumps from
59.5% to 74.2%. This difference is statistically significant at a 1% level (Mann-Whitney
test, p-value 0.0067). This difference is also evident when comparing average
cooperation in the first period of each cycle (73.5% vs. 84.5%, Table 4). Surprisingly,
average cooperation is statistically indistinguishable from the non-anonymous public
monitoring treatment (Mann-Whitney test, p-value 0.154).
Result 5 (Realized efficiency). In the anonymous treatments, the introduction of
personal punishment increased realized efficiency over private monitoring and over
public monitoring.
The comparison among treatments in terms of realized efficiency substantially
confirms the conclusions drawn in Results 1-4 in terms of average cooperation. We
1 Tk 4
define realized efficiency in an economy k as ek = — ∑ ∑ (πk -1θ) / (25 -10).
T t=1 i =1
21