Result 2. Cooperation did emerge in economies with private monitoring. Cooperation in
later cycles is higher than in earlier cycles.
Figures 2 and 3 and Table 4 provide support for Result 2. In the private monitoring
treatment average cooperation across economies was 59.5% for all periods and 73.5% for
just the first periods, which are remarkably high given results in previous studies
(Schwartz et al., 1999; Ochs and Duffy, 2006). This provides support to the empirical
relevance of the theoretical results of Kandori (1992) and Ellison (1994).
The average cooperation had an increasing trend across cycles, as seen in Figure 4.
100%
<л
с
о
Ü
га
а>
>
≡
CU
о.
о
о
и
о
C
о
Ü
га
Figure 4: Average cooperation across cycles
- - Private Monitoring With Punishment
—N—Non-anonymous Public Monitoring
80%
60%
40%
20%
0%
cycle 1 cycle 2 cycle 3 cycle 4 cycle 5
This figure suggests that as subjects became familiar with the incentive structure of
the indefinite repetition, they responded by increasing cooperation level.12 This finding is
in line with previous studies (Aoyagi and Frechette, 2003; Dal Bo and Frechette, 2006)
and marks a difference with finitely repeated prisoners’ dilemmas and voluntary public
12 There is statistically significant learning across cycles. We aggregated economies from all treatments by
cycle and carried out Mann-Whitney tests of pairwise differences in cooperation between cycles. The
increment between cycle 1 and 5 is significant at 1% level. The most significant jump in cooperation level
is from cycle 1 to cycle 2 (5% significance) while the difference between cycles 4 and 5 is not significant at
a 10% level (one economy is one observation; in each comparison n1=n2=40).
20