average number of times in which the given punishment was applied in the past year is also reported.
In the full sample, the percentage of groups imposing a penalty for missed contributions is 38%, and
that for irregular behavior is 47%. When these statistics are computed separately for ‘high’ and
‘low’ heterogeneity groups, we find that only 14% of the ethnically heterogeneous groups penalize
missed contributions, while 67% of the homogeneous ones do. Wealth inequality works in the same
direction, though the discrepancy is somewhat smaller. As for education heterogeneity, the pattern is
reversed but we have already seen that this index seems to capture skill complementarities more than
‘conflict’. From the number of times in which the penalty is applied, it seems that when ethnically
fragmented groups do envisage a penalty, they apply it more often than others (though this may be
due to higher need for punishment due to increased tension, as opposed to greater effectiveness).11
Sanctions for absenteeism and irregular behavior seem more or less equally likely and uniformly
applied between high and low heterogeneity groups, suggesting that the effect of fragmentation may
be closely related to control over economic resources.
Finally, the last panel of table 7 explores whether there is a relationship between the criteria
for recruitment and the composition of the group. For example, if new members are chosen on
the basis of their ability we may expect groups to be more heterogenous than if they are chosen
because they are ‘friends’ of other members. In the aggregate, group leaders say that the single
most important criterion to select new members is ‘commitment’ (50% of the groups), indicating the
degree of uncertainty faced in their everyday work. The remaining criteria are ability (22%), need
of the member (22%), and only in rare cases friendship (5.6%). Interestingly, no clear relationship
emerges between the recruitment criterion and the heterogeneity of the group, which may be due to
the limited importance of the ‘friendship’ factor in determining who joins. When individual members
were interviewed about the main reason for joining the group, only 6% said that they had friends in
the group. All this is comforting because it is consistent with the interpretation that the causal link
goes from heterogeneity to decision making and group performance, and not the opposite. In fact,
11The apparent inconsistency between the presence of a penalty for missed contributions and the number of times
in which it is applied may be due to measurement error, but it may also indicate that, even when in place, sanctions
in heterogeneous groups do not deter deviations, while in homogenous ones they serve a pre-emptive role.
18