landowners pressing for proportionally more of the irrigation supply. The Mexican study also finds that proportional water allocation is associated with poorer maintenance. Inequality thus may lead to a particular set of rules under which irrigation systems do not perform as well.²³ Bardhan’s (2000) south Indian study also provides significant evidence that when the water allocation rules are crafted by the village elite, the latter violate the rules less frequently; otherwise the elite violate the rules more frequently. (Overall, the elite break the rules more often than the nonelite, but by definition, the former is a smaller group than the latter.) It is also observed that when an average farmer believes the water rules have been crafted jointly (i.e., with collective participation, as opposed to rule crafting only by the village elite, or by government), he is more likely to have positive comments about the water allocation system and about rule compliance by other farmers. Joint rule crafting of this kind is associated with the highest level of rule compliance in the Indian case. Bardhan also estimates the likelihood that villages have adopted proportional cost-sharing rules—that is, rules that specify that the labor costs of maintaining irrigation infrastructure are shared proportionally to (irrigated and nonirrigated) landholding wealth.²⁴ This rule is in general positively associated with cooperative outcomes; adoption of this rule is, in turn, significantly and positively associated with landholding inequality. This might be an indication of social pressure for a redistributive adjustment of the cost-sharing rule to take account of wealth disparities. This points to an important and more general observation noted by Varughese and Ostrom (2001:762). They find that many groups “overcome stressful heterogeneities by crafting innovative institutional arrangements well-matched to their local circumstances.” In their Nepal study, forest users created diverse forms of memberships with different rights and duties to cope with heterogeneity, particularly when there are substantial benefits to be obtained through collective action.

Table 3-1 summarizes the evidence from the large-n studies reviewed in the previous section. The evidence is tentative, but sufficient to permit us to hazard a few conclusions. First, there is a confirmation of the case study literature’s conclusion that heterogeneity, however it might be measured, has a negative impact on cooperation on the commons in these irrigation cases. Heterogeneity tends to have a discernable negative effect, or no effect at all. Second, the evidence is consistent with the hypothesis that heterogeneity weakens the effect of social norms and sanctions to enforce cooperative behavior and collective agreements. Support for this conclusion derives from the negative effect on performance in multivillage, multicaste irrigation systems. Third, however, controlling for this social heterogeneity, there is an independent, and largely negative, effect of economic heterogeneity per se: This is borne out by the significant effect of Gini