coefficients, for example, in the Mexican and Indian studies. This conclusion underscores the importance of economic mechanisms in the theoretical literature, starting with Olson (1965). These economic mechanisms are based on the differential incentives to cooperate created by the distribution of wealth or income, distinct from social norms. Moreover, although economic theory cannot predict whether “Olson effects”—a positive impact of inequality—will predominate, the empirical evidence for irrigators is that they do not. This finding also underscores the value of the multivariate analysis approach adopted in the studies summarized here: Such an approach allows one to isolate the effect of particular structural characteristics (like wealth inequality) while controlling for the effect of others (like social heterogeneity). Fourth, and finally, there is evidence that heterogeneity affects system performance both directly and indirectly via its effect on the institutions adopted by an irrigating community. Inequality might affect the degree to which irrigators follow the rules, but it also affects the type of rules chosen, and not all rules are equally conducive to good performance. Quantifying the magnitude of these direct and indirect effects requires the adoption of the multivariate approach used in these studies.
A question raised by the studies summarized here is the degree to which these results based on irrigation systems can be generalized to other types of commons. Blomquist et al. (1994) consider a typology of common-pool resources that situates irrigation with respect to other types of commons. Two physical dimensions that matter are stationarity of the resource (“the resource units… remain spatially confined prior to harvest” [1994:308]) and the possibility of storage. This two-way typology generates four classes of physical resources, and irrigation systems are found in three of the four categories: groundwater-based systems are stationary and storage is available; reservoir-based canal systems are nonstationary but storage is possible; river-diversion systems are nonstationary and storage is impossible. Many of the systems considered in this chapter are canal-based systems, with a minority of run-of-the-river systems. (In Bardhan’s 2000 South Indian study, half of the systems are not served by canals.) Most other commons do not share this combination of characteristics: Forests, rangelands, and community threshing grounds are stationary without recourse to storage. Migratory species are nonstationary with no possibility for storage. In the presence of distinct structural characteristics, further analysis is necessary before extending these irrigation-based findings to other settings.
One distinction between a flowing resource (like water) and a standing resource (like forest) is that gravity makes locational (head-tail) heterogeneity more salient in the former (although if land markets work, one presumes that in the long run locational advantage gets capitalized into wealth inequality). Another difference is that in forests, part of the collective action is in replanting and regeneration efforts, which are less important in canal irrigation. Third, issues of intertemporal conservation are somewhat less salient in canal irrigation, although