This paper attempts to explain why collective organizations in charge of managing scarce resources can face difficulties in evolving into more efficient institutions. It is framed within a literature strand that asks why institutions fail or succeed, on the basis of some skepticism about the use of rational models of social design as well as a deliberate openness to learning from cultural and biological evolution. Rational order, in the words of Vernon Smith (2003), can be thought of "as an undesigned ecological system that emerges out of cultural and biological evolutionary process: homegrown principles of action, norms, traditions and 'morality'" (469-470). However, R. Maria Saleth and Ariel Dinar (2004) have argued, for the case of the water sector, that institutions are severely constrained by path dependency, a construct that allows them to illustrate the powerful effects of history, customs, and traditions on the observed state of water institutions around the world. Saleth and Dinar concluded that path dependency poses serious constraints for change and evolution.
Some authors have built on Elinor Ostrom's works to analyze water user associations around the world (Ostrom 1990, 1993, 2001; Livingston 1985; Subramanian et al. 1997). This literature teaches us how collective entities share costs, risks, and benefits. It also shows what modes of collaboration lead to more social cohesion and to more efficient ways of dealing with various sources of collective and individual risks. Uncertain water supply may be one of the most important for irrigation schemes but certainly not the only one.
Constructivist rationality, quoting Smith (2003) again, stands opposed to evolutionary biology in that it "uses reason to deliberately create rules of actions, and create human socioeconomic institutions, that yield outcomes deemed preferable.... Although constructivism is one of the crowning achievements of human intellect, it is important to remain sensitive to the fact that human institutions and most decision making is not guided primarily, if at all, by constructivism" (468). In the context of water resources and collective organizations, rational constructivism leads to the notions of property rights, incentives, and resource tradability. This mindset supports many sectoral reforms that give markets a major role in resource allocation.
Yet a plain fact encountered in many worldwide irrigation organizations is that formal water trading is an exception, whereas one can find numerous ways of informal trading in most countries where irrigation water is a basic resource for food production. Furthermore, while virtually all world countries have formal water codes, very few recognize that water rights can be tradable (Australia, USA, Chile, and, since 1999, Spain; see Easter et al. 1998), and most of them explicitly forbid water users to exchange their water rights. This finding is somewhat disturbing, as one would expect that states would be inclined to regulate the markets and develop means to make them more secure, equitable, and acceptable.
We argue that both Smith's ecological rationality and Saleth and Dinar's path dependency offer alternative, but not exclusive, explanations of the weak support for formal water trading within irrigation communities. In a context of uncertain water supply, risk analyses offer opportunities to understand the inherent difficulties of establishing formal trading rules. In addition, users' heterogeneity, stemming from unequal endowments, preferences, or culture, reduces the chances of forming a commonly agreed-upon understanding of the resource's allocation problems. Yet organizations develop ways to cope effectively with users' heterogeneity, which in our case is illustrated by the evolution from pure proportional allocation systems to customary priority systems that reward risk-taking irrigators.
This paper seeks to explain why it is not easy for water associations to adopt water-trading rules. We show that collective institutions may ecologically evolve along paths that can hardly lead to acceptable trading rules, approved through democratic voting mechanisms. In our case, unstable water supplies have been faced with collectively readjusting a purely proportional distributional doctrine, which is sanctioned by most water codes, into one in which the priority criterion ensures that farmers with the most valuable crops enjoy preferential access. Thus, the system has evolved along a path that features two distinctive elements. First, it has transformed the formal water rights into customary pseudo-rights based on priority rules that are not backed by any other means than customs. Second, it ensures that scarce water is allocated to the most productive and risk-taking users, though one can surmise that the customization of rights and the investments made by the most productive users have evolved in parallel, feeding each other.
We take on the example of Spain for a number of reasons. First, there are about 6,000 irrigation communities, many of them centuries old. Second, Spain has an advanced water code judged on world standards, and formal water rights even before the Roman Empire ruled the Iberian peninsula, more than two millenniums ago. Third, irrigation communities have formal and customary rules and recorded data, and they follow democratic voting procedures for decision making. Fourth, evidence of potential gains from intrasector water trade is at odds with the evidence that shows that farmers are reluctant to trade in formal markets (Garrido et al. 1996) or may require a number of years to get used to it once they are established (Bjorlund 2003). One final reason is irrigation communities follow democratic voting mechanisms to change their rules and statutes.
We base our arguments on the results of a modeling exercise that includes various risk measures of individual farmers' profits evaluated for four possible market and nonmarket scenarios. Generally, the literature on water markets does not take the analyses beyond the evaluation of social and individual economic efficiency gains. Yet in this paper we take these individual orderings and, based on the voting structure governing the irrigators association, obtain hypothetical votes for various water market and rights reforms. Underlying our modeling framework is the fact that water supply for irrigation is uncertain for climatic reasons. This explains our emphasis in profit risk measures rather than expected outcomes.
The voting mechanisms, individual farmers' representativeness, and the likely profits resulting from each scenario suggest that finding a sufficient majority for establishing a formal market is very difficult. Thus, although any alternative market design would lead to more efficient water allocation, none may be acceptable for a majority of users. We show that the transition from nontradable water rights, adjusted by risk reduction customs, to tradable water rights may be hindered by the "cycling" effect. As Michael Rosen and Richard Sexton (1993) explained, cycling occurs when a collective organization's preferences over several alternatives are not transitive even though each member's preferences may be so.
A Description of the Institutional Setting
Water Allocation Mechanisms
In the absence of formal water trading rules, one commonly finds irrigation communities that ecologically evolve to face the risk derived from uncertain supply, developing creative institutions. An example of this is the allocation of scarce water favoring farmers with permanent crops and horticultural crops. The literature on collective irrigation organizations suggests that customs are forerunners of formal rules and statutes. In the case of Spain, irrigator communities have framed their allocation customs in a manner that favors risk takers, although most often water rights are not formally redistributed to accommodate these preferential uses. Rather, irrigation communities have typically set out to internally reassign firm proportional water rights in order to favor some users in times of scarcity. They conform with Ostrom's judgment (2001) that "in a self-governed system, participants make many, but not necessarily all, rules that affect the sustainability of resource system and its use" (1317).
In a stylized form, any allocation criteria can be represented by
[W.sup.i.sub.t] = min ([W.sub.i] [f.sub.i]([W.sup.S.sub.t]) x [W.sup.S.sub.t]); with [f.sub.i] ([W.sup.S.sub.t]) [less than or equal to] 1, [for all]i
where [W.sup.i.sub.t] denotes the amount of water that holder i is entitled to during period t; [W.sub.i] is the maximum amount he can use, as stated in his water right; [W.sup.S.sub.t] is the known stock of water that the basin agency decides to make available to the whole community during period t; [f.sub.i]([W.sup.S.sub.t] is one way to represent the priorities made explicit in the bylaws. This proportion is specific for each category of users. If user j is junior to i-a farmer with permanent crops--then [f.sub.j]([W.sup.S.sub.t]
The rationale of preferential or priority sharing rules is clearly to protect the permanent crops and, sometimes, to favor small farmers or the plots that are planted with horticultural crops. In Andalusia (South of Spain), farmers growing horticultural crops--called "social crops" because they demand more labor--receive a larger proportion of water than those that plant extensive crops (cereals and protein crops). Commonly, small holders complement their household income working as skilled laborers in larger horticultural and fruit tree orchards.
A related issue is the origin of the criteria that give rise to an asymmetric distribution of scarce water resources. Here the evidence is dispersed and indicative of the importance of the historical processes. Rutgerd Boelens...