who uses project level data on 132 community-maintained infrastructure projects in the Himalayas.
The author finds that, ceteris paribus, socially heterogeneous communities have poorly maintained
projects, and that the relationship between maintenance and inequality in projects returns is U-
shaped. Miguel (2000) and Gugerty and Miguel (2001) analyze the effect of ethnic diversity on school
funding and school organization in Kenyan villages. Both studies find that ethnic fragmentation is
associated with worse outcomes, and suggest mechanisms though which this effect may be generated.
In particular, Gugerty and Miguel find that the more racially heterogeneous the pupils’ population,
the lower are parental participation in school activities, school committee attendance, and teacher
attendance and motivation. The authors suggest that these results may be due to the fact that
social sanctions apply within ethnic groups, hence fragmented communities have lower scope for
sanctioning noncooperative behavior. In a recent study, Karlan (2001) examines the impact of
members’ composition on savings and repayment performance in group lending programs offered by
a Peruvian organization. The author focuses on geographic and ‘cultural’ dispersion (Western versus
indigenous) andfinds that both types of heterogeneity increase the probability of defaulting on group
loans. As for savings, which can be viewed as a form of ‘collective’ good to the extent that capital is
lent back to the members, geographic distance among the members significantly reduces saving rates,
while cultural dissimilarity has a negligible effect. Finally, Wydick (1999) estimates the determinants
of repayment performance and of intra-group insurance in a sample of 137 borrowing groups from
Guatemala, and finds that ‘social ties’ (as measured by the number of years members have known
each other and by the degree of friendship and social interaction before and after joining the group)
have a negligible impact on group performance. What seems to matter most is the availability of
sanctioning mechanisms and the possibility of monitoring (as proxied by geographical distance and
knowledge of each other’s business).
The present study builds upon the above literature with a few significant modifications. First,
it considers groups that provide the main source of income for most of the members. In other
words, individuals’ stake in these groups is particularly high in this case as compared to project
maintenance groups or school committees, and possibly even to borrowing groups, in that members
mostly depend on production activities organized within these groups for their daily living. As such,