Appendix B
Table 5: Probit regression on individual choice to cooperate - marginal effects(*)
|
Dependent variable: |
Private |
Anonymous Monitoring |
Private |
Public |
All treatments |
All |
|
Treatment | ||||||
|
dummies: | ||||||
|
Anonymous Public |
-0.04566* |
-0.02924 | ||||
|
Monitoring |
(0.02450) |
(0.07313) | ||||
|
Private Monitoring |
0.99813*** |
0.09188 | ||||
|
With punishment |
(0.00010) |
(0.06737) | ||||
|
Public Monitoring |
0.94668*** |
0.11747* | ||||
|
(non-anonymous) |
(0.00951) |
(0.06082) | ||||
|
Cycle dummies: |
0.03938 |
0.05750 |
0.08336*** |
-0.00304* |
0.06190*** |
-0.03677 |
|
Cycle 2 | ||||||
|
(0.10412) |
(0.03787) |
(0.02618) |
(0.00179) |
(0.02336) |
(0.02852) | |
|
Cycle 3 |
0.07621 |
0.05045 |
0.11154*** |
0.02014*** |
0.09321*** |
0.00613 |
|
(0.06949) |
(0.05102) |
(0.01972) |
(0.00188) |
(0.02688) |
(0.02911) | |
|
Cycle 4 |
0.13560*** |
0.18856*** |
0.14888*** |
0.12585*** |
0.17362*** |
0.04925 |
|
(0.00770) |
(0.02515) |
(0.03042) |
(0.02706) |
(0.02205) |
(0.03559) | |
|
Cycle 5 |
-0.16040*** |
0.28969*** |
0.13911*** |
0.13946*** |
0.21443*** |
0.08162*** |
|
(0.04322) |
(0.03196) |
(0.03295) |
(0.00420) |
(0.02114) |
(0.03126) | |
|
Duration of previous |
0.00133* |
0.00272*** |
0.00249*** |
0.00413*** |
0.00362*** |
0.00327*** |
|
cycle | ||||||
|
(0.00076) |
(0.00015) |
(0.00033) |
(0.00058) |
(0.00082) |
(0.00073) | |
|
Reactive strategies: | ||||||
|
Grim trigger |
-0.55054*** |
-0.26642*** |
-0.38185*** |
0.07527 |
-0.38854*** | |
|
(0.01458) |
(0.07434) |
(0.10006) |
(0.05562) |
(0.04064) | ||
|
Tit-for-tat with lag 1 |
0.08814** |
-0.04836** |
0.05643* |
-0.06063 |
0.01797 | |
|
(0.04303) |
(0.02415) |
(0.02986) |
(0.03933) |
(0.02747) | ||
|
Tit-for-tat with lag 2 |
0.11608*** |
-0.09505*** |
0.04598* |
-0.14011*** |
-0.02680 | |
|
(0.03620) |
(0.01776) |
(0.02692) |
(0.03122) |
(0.03870) | ||
|
Tit-for-tat with lag 3 |
0.10324** |
-0.07310* |
0.04044 |
-0.06265*** |
-0.00994 | |
|
(0.04237) |
(0.04174) |
(0.03453) |
(0.00683) |
(0.02686) | ||
|
Tit-for-tat with lag 4 |
0.08009*** |
-0.05838 |
0.01518 |
-0.05327 |
-0.03325 | |
|
(0.00524) |
(0.04672) |
(0.04463) |
(0.06054) |
(0.02879) | ||
|
Tit-for-tat with lag 5 |
0.02974** |
-0.07094*** |
0.01432 |
-0.01770 |
-0.04448* | |
|
(0.01385) |
(0.00729) |
(0.03021) |
(0.04159) |
(0.02311) |
(*) Marginal effects are computed at the mean value of regressors. Robust standard errors for the marginal
effects are in parentheses computed with a cluster on each session; * significant at 10%; ** significant at
5%; *** significant at 1%. For a continuous variable the marginal effect measures the change in the
likelihood to cooperate for an infinitesimal change of the independent variable. For a dummy variable the
marginal effect measures the change in the likelihood to cooperate for a discrete change of the dummy
variable.
First periods of each cycle are excluded with the exception of the last column. Individual fixed effects and
period fixed effects are included but not reported in the table (individual dummies: s2-s30 s32-s37 s39 s41-
s60 s62-s97 s99-s159; period dummies: 3, 4, 5, 6-10, 11-20, 21-30, >30). A grim trigger regressor has
value 1 in all periods following a defection and 0 otherwise. Five tit-for-tat regressors have value of 1 only
in one period following a defection and 0 otherwise; we trace response up to a five-period delay. Details on
strategy coding are in the text after Results 7. Duration of previous cycle was set to 20 for cycle 1.
40