Résumé
Despite the fact that agriculture will have to adapt to face increasingly frequent and more severe droughts with climate change, existing water allocation mechanisms (pricing, water markets, rationing scheme) often fail to take into account the high variability of the resource. Water allocation schemes can contribute to reduce the cost of drought risk for farmers and society by integrating risk management mechanisms. We address this question in two different contexts: France, where water is managed by public authorities that are responsible for the rationing scheme and defining the allocation of water between farmers in case of scarcity; Australia, where water markets have been implemented in order to allow water transfers between farmers. This Ph.D. dissertation is composed of 3 articles (chapter 2, 3 and 5 in English) and background information on the organization of irrigation water allocation in France (chapter 1 in French) and the allocation of irrigation water in case of drought in Australia (chapter 4 in French). Chapters 2 and 3 analyze the interactions between the restrictions rules used in France to share the resource in crisis times and the self-insurance strategies of farmers. We investigate the particular example of on-farm storage. Chapter 2 models the decisions of agents heterogeneous in their risk tolerance, who can substitute the use of a secure private alternative to the use of a common pool resource of uncertain size. We find the optimal sharing rule under two assumptions concerning the regulator: i) the social planner defines simultaneously the sharing rule and the diversification choices in order to maximize social welfare; ii) the policy maker uses the sharing rule as an instrument to induce each agent to make the optimal diversification decision. Chapter 3 uses laboratory experiment to analyze how the sharing rule impacts the individual trade-o_ between relying on a free but uncertain resource and investing in a secure alternative resource. Rules are compared on the basis of their ability to facilitate coordination on the optimal investment in the secure resource. Chapter 5 studies with a laboratory experiment, designed to capture the main characteristics of Australian water markets, whether offering different levels of security for water rights could potentially allow water users to manage the risks of supply uncertainty better. We find that a two security levels system improves risk sharing between users. However, a system with different levels of security for water rights can be detrimental to cost-effectiveness by increasing the number of costly trades on the water rights market, when transactions costs are higher in this market than in the market for allocation.