shows that subjective inequality is positively correlated with the extent of inequality in income
received by different members from the group (the correlation between the two is .64). This result
can be rationalized by presuming that people have limited information about each other’s personal
belongings, but that they have fairly accurate information about how much each member is paid,
which matches the environmental conditions of the informal settlements and the groups surveyed
pretty closely. If subjective beliefs are formed this way, then, it is not surprising that inequality
in earnings affects group functioning more than wealth inequality: members’ ability to agree on
collective outcomes may depend more on their beliefs than on their objective conditions.
The bottom panel in figure 1 plots the fraction of people who think that members differ a lot
in wealth against the likelihood that two randomly drawn members of the group belong to different
ethnicities. The link between ethnicity and wealth is not obvious in this context, but it may be
that simply by considering a member as an “outsider” leads an individual to label him or her as
very different in terms of wealth. The figure suggests a weak positive association between the two
dimensions (the correlation coefficient is .10). Finally, in no case the heterogeneity indexes considered
vary systematically with size, which means that higher heterogeneity does not simply capture a larger
group of participants (which may be considered a source of collective action problems).
[Insert table 7]
In the light of the above discussion it is useful to examine the impact of various dimensions of
heterogeneity on the organization of production within the group. Thefirst column of table 7 reports
summary statistics on the percentage of groups that fall into different forms of organization: this pro-
vides a broad idea of the way these groups function. The same statistics are then computed splitting
the groups into ‘High’ and ‘Low’ heterogeneity depending on whether their degree of heterogeneity
is above the median or not (where the median is the index of heterogeneity of the median group).
The dimensions of heterogeneity considered are ethnic, in wealth, and in education. Inequality in
earnings from the group is omitted because it is endogenous to the organization of production chosen.
Comparing the incidence of the various organizational forms in high and low heterogeneity groups
can be a first step towards understanding how heterogeneity affects group functioning, though such
16