Abstract : This thesis proposes an analysis of food price stabilisation policies in poor countries. In order to represent the conditions in these countries, we introduce in a rational expectations storage model the assumption that consumers cannot insure against price risk and are risk averse. The market incompleteness justifies public intervention and various stabilisation policies are analysed. An optimal storage policy implies a storage level higher than without intervention. This additional storage crowds out private storers by removing the profit opportunity from speculation. Since the use of complex intervention rules is unlikely in poor countries, we compare state-contingent optimal food storage policies, with and without commitment, to simple storage rules such as a constant private storage subsidy or a price-band defended by public storage. The ability of government to commit to a policy rule brings additional welfare gains in comparison with a discretionary policy. These gains stem from the possibility of manipulating producers' expectations and, thus, of inducing them to promote stabilisation. Simple stabilisation rules can achieve, when designed optimally, gains closed to those of state-contingent policies. Finally, the analysis of stabilisation policies is extended to an open economy framework in which the stabilisation instruments are trade and storage policies. An optimal storage policy alone fails to protect consumers, since most additional storage is actually used to serve the world market. In contrast, an optimal trade policy strongly decreases price volatility by exploiting the world market.