Worldwide, food reserves have regained interest following free market incapacity to handle 2007-08 world food crisis. The issue in the implementation of food reserve policies is to account for countries specific context. In most West-African countries, and Benin specifically, few or no empirical researches have been conducted at macro and/or micro levels on the appropriateness and effectiveness of the current grain reserve policy. Benin grain reserve policy includes two programs: 1) market access program in which farmers are beneficiaries. It includes a procurement system coupled with an information system; 2) price subsidy program: fair-price-shop program where consumers are beneficiaries. To analyse the policy appropriateness, Stated Preference Discrete Choice (SPDC) Experiment modelling was used focusing on the procurement system. And to evaluate the policy effectiveness a novel Multivariate Endogenous Treatment Effects (METE) model, developed in Structural Equation Modelling (SEM) frame was used to accommodate for the policy simultaneous effects on stakeholders diverse welfare indicators. METE model could be used in a situation where a treatment (e.g. policy) affects simultaneously treated agents multiple welfare indicators. These tools were used based on data collected from each stakeholders group (i.e. farmers, traders and consumers). The choice experiment modelling reveals that the important attributes for traders participation in the market-based procurement system are: better coordinated supply chain, provision of incentives to participants and shorter payment delays. And from hypothetical procurement designs evaluated, the most appropriate is a supply chain including three actors: assemblers, small-wholesalers and wholesalers; coupled with 10 days of payment delay and provision of weekly price information as incentive. The impact evaluation approach shows that among beneficiaries, only farmers are worse off. To deliver positive impacts on farmers, the study suggests: a) at institutional level, to improve both procurement and information systems designs and b) at farm households level, to train farmers on own-produced cereal (i.e. maize) allocation methods in favour of less unbalanced sharing between consumption and market demand; and to promote staple food stocks building to increase households resilience to price fluctuations.